A dark galaxy by any other name

Back in February I got very excited by the discovery of a really good, solid, dark galaxy candidate courtesy of China's FAST telescope. I haven't heard anything more about that since then so I remain excited about it.

Today's paper also comes via the FAST telescope and concerns another dark galaxy candidate. However this isn't by the same authors. It's actually based on this paper from June (which I've not read) by the FAST team, but the authors are unconnected. Basically what they're doing is saying, "hey, the FAST paper said that this might be a dark galaxy, and here's our more detailed analysis of the same data supporting that". There isn't any new data in this paper, just pure analysis. In fact I'm not sure if they even have access to the FAST data (which I would have thought was private) or just scraped it from the published images... 

Look, dark galaxies are the thing that gets me most excited about astronomy, but I have to say the more I think about it, the less I'm a fan of this particular paper. It's not that there's necessarily anything wrong with it, it's just... odd, as a piece of work.

It's not just the lack of new data and their unclear access to the original data* that irk me about this paper. They also don't cite me, even though my work would be directly relevant here. And they insist on calling "dark galaxies" RELHICs ("RE-ionisation-Limited HI Clouds").... urrgh, that's an ugly term. Most irritating of all, they use the term "isocontour". The hell is that ? Are normal contours not plotting things at the same (iso) value anyway ?

* If they had it, surely they could have reprocessed it in a few other ways and at least reported what they did, even if it didn't show anything new ? Not to do this is very strange.  

To be fair, RELHIC might be better as a term were it reserved for a very specific kind of dark galaxy : low HI mass, low rotation speed, small dark matter halos, as in this putative case. I'm not sure we really need another term for this besides the already-common "minihalo", but still.

Anyway what they do is to show that this particular object is consistent with being a RELHIC/minihalo/dark galaxy, in which the HI is in hydrostatic equilibrium. It would need a dark matter halo to be stable because the HI mass is undoubtedly far too low to keep it self-bound. And it can't be much further away or closer than the best-guess distance estimates, because that would make it an altogether stranger object.

And that's pretty much it. Even this depends quite a lot on the shape of the HI contours, and I'm a bit concerned that this circular shape - the distinguishing feature of a non-rotating system in equilibrium - is only due to the low resolution of FAST (though as they say, higher resolution would be better). All this is extrapolating quite a lot from very marginal data - I don't doubt the detection is real, only the confidence about the conclusions. 

Nor do they consider other explanations either. True, it's quite well-separated from the nearest big spiral, M94, but 70 kpc is not that far. And as we and others have shown, objects with line widths this narrow (a mere 20 km/s, only twice the width of the line itself) can be stable on very long timescales even if they're completely unbound, because by definition they're dispersing slowly. 

In short, the paper presents a perfectly valid summary of what the object could be, but it doesn't offer anything new and doesn't consider alternatives. I'd have been far happier about it if they'd at least applied this analysis to other objects and/or used at least some other data sets here. On the other hand, at least dark galaxies are getting more attention again, and that at least is something to be welcomed.

Lights in the darkness

At long last I rouse myself from a prolonged refusal to read any papers with this one about molecular gas in the Leo Ring.

The Ring, you might remember, is a giant atomic hydrogen (HI) structure over 200 kpc in diameter whose origins remain unclear. It's of particular interest to me for two reasons. First, we have AGES data of the Ring which found a set of small, discrete clouds nearby that may or may not shed light on its origins. Second, we might be able to get time with the APEX telescope to look for molecular gas here, so today's paper was some much-needed background reading on my part.

Actually many of the authors involved in this paper were also involved in this one in 2021, which found several small patches of star formation happening in the densest parts of the Ring. Here they follow-up those patches using the IRAM 30 m radio telescope to look for molecular gas. Star formation seems to be reasonably well-behaved in typical galaxies, but nobody is quite sure if this is because of the fundamental physics of gas collapse or only because the conditions within galaxies tend to be similar. Looking at how it works in the Ring would be a great way to test this, because here there's little or no rotation and the gas density is much lower, yet the chemical composition is similar to ordinary galaxies.

I have to say I found parts of this paper a bit of a slog and some of the narrative could have been clearer. As far as I can tell, they did 11 different pointings in and around two Halpha (ionised gas) regions, but their choice of nomenclature is horrendous. They also stack the individual observations to give one equivalent to 72 hours of observing time, but don't seem to say how long each individual observation took. 

The basic result is that they don't detect anything. They have some hints of marginal detections, but with commendable honesty, they amply stress just how tentative these are. For example some of the "detections" have mismatched velocities at different frequencies, and the stack doesn't reveal a detection. Additionally, given their velocity widths the lines are brighter than the usual scaling relations suggest. All things considered, it's quite probable that they're all spurious.

Fortunately this doesn't matter too much. Their sensitivity is high enough that this becomes a genuinely interesting non-detection exercise; based on the observed star formation rates, typical clouds should have been massive enough to be detectable. So why aren't they ?

There are several possibilities. It could be that the physics of star formation is indeed different in this environment. It might be so efficient that much lower mass clouds are able to form stars, and so rapid that the onset of star formation quickly suppresses any further collapse (e.g. by stellar winds and supernovae, which would disperse the gas on local scales though would have no affect at all on the much larger scales of the Ring). Interestingly they say this could explain Ultra Diffuse Galaxies, though I suppose we'll have to wait for metallicity measurements of those objects to see if they're comparable with the patchy star formation in the Ring.

There are other options. It could be that the star-forming gas here is dominated by atomic rather than molecular gas, an idea that's been floating around for a while. Alternatively it could be that the CO they observe here isn't such a good tracer of molecular hydrogen as in normal galaxies (so-called "CO dark" molecular gas), an idea that's extremely popular with theorists. Or perhaps there is molecular gas but it's a little further away from the star-forming regions than usual, just beyond where the telescope was pointed.

Bottom line ? Something is different about star formation in the Ring than in galaxies. That's progress. I also have to give them high praise for describing their sensitivity estimates in considerable detail, and while of course detections always give you more to work with, they seem to have thoroughly exploited what they've got. But to really say what's going on here, I think needs even deeper observations at different wavelengths. Constraints can only get you so far, at some point, you need to see something.

When Is Something Massive But Not Massive ?

... when it's a massive galaxy that's not as massive as it should be.

What's that you say ? It's been more than five minutes since you heard about galaxies without any pesky dark matter, and you're desperate for a fix ? Look no further !

To briefly summarise : it looks as though the two candidates which started the ball rolling in this field really do lack dark matter, but a plausible, specific collisional formation mechanism can be invoked to explain them. The stellar masses of these objects are very low, as are those of the other major candidates... though those particular (ultra diffuse) galaxies are harder to explain because they're much more isolated, and additionally only seem to be partially deficient in dark matter rather than lacking it entirely.

The latest candidate, the subject of today's post, is very different to the others. It's much brighter, with a stellar mass three or four orders of magnitude greater. That immediately raises some flags that maybe this is altogether a stranger and even more interesting object : in general, you can do anything you like as long as your galaxy is pathetic enough, but the bright ones tend to fight back. If they're doing something weird, you're on much firmer footing in claiming to have found something seriously strange.

There's a press release here which is quite nice. Of course, I tried to read as much as the original paper as I could, but I have to say I struggled with this one. Large parts are extremely dry and technical and read more of an instruction manual than a scientific analysis. It's absolutely no bad thing to go into this level of detail - in fact, authors, I salute you for your efforts ! - but I do think it could have been structured more effectively. An awful lot of this belongs in an appendix.

I also have to say I found some of the wording... odd. Not linguistically odd at all, just arranged weirdly, jumping from topic to topic, flitting between technical details of the procedures and the scientific analysis, and occasionally coming across as though they meant to say the opposite of what they seemed to be claiming. Narratively, to be honest I just don't like it, it doesn't flow. So after wading though the methodology as far as I could, I skipped ahead to the discussion section, which is it least comprehensible even if it's a bit off.

But let's start with at the beginning. Here they introduce the current prevailing view on the formation of elliptical galaxies, which I'm not all that familiar with. As per much earlier suggestions, it looks like both of the major paradigms of galaxy formation theory (monolithic collapse, suggested in the 1960s, and the much more recent idea of hierarchical merging) might be at work for these objects. Monolithic collapse - the simple collapse of a great big singular cloud of gas - creates an initial, highly compact "nugget", which then grows and expands over time through mergers with other (gas poor*) galaxies. Since those objects lack gas which can dissipate the energy of the collision, the nugget expands spheroidally rather than forming a rotating disc.

* If instead it encounters gas-rich galaxies, presumably this would assemble a spiral/disc galaxy.

There's no guarantee that a nugget must experience this second phase of merger-based growth though. Oh, most will, just because galaxies are belligerent and numerous. But in rare cases it's possible that they don't, and the result of the nuggets evolving passively is a very rare "relic galaxy", of which only a few dozen candidates are known.

The second really interesting point in the introduction kindof makes the rest of the paper feel like a bit of a let-down : that this particular candidate relic galaxy, NGC 1277, is already known (since 2015) to lack dark matter out to a distance of 4.6 kpc from its centre. This study extends that to... 6 kpc. No matter how much the authors try, it's a bit difficult to get excited by this undeniably important yet also undeniably incremental change. Fortunately for them, since I never heard about this before, it's still a very interesting object. (And it's pretty neat that we're still making discoveries like this from NGC objects, a catalogue which is more than a 130 years old ! This is quite the reminder of our sheer ignorance.)

The new observations measure the stellar velocities over a field which roughly corresponds to the following :

NGC 1277 is upper right, while the larger galaxy near the centre is NGC 1278.

Zooming out, this is a crowded part of the Perseus cluster. This causes some complications, but the obvious advantage is that they get two galaxies for the price of one.

What they find is that the motions of the stars of NGC 1277 are consistent with what's expected from the mass of the stars alone (i.e. slow), whereas those in NGC 1278 are what you'd expect if the galaxy also had the usual amount of dark matter (fast). These measurements seem pretty robust, so it's not at all likely that people will dispute the accuracy of the data, unlike in some of the earlier cases.

Straight off, this is bad but not fatal news for dark matter alternatives. Dark and normal matter are in principle separable - there's no reason that the one must accompany the other. We've seen this more famously in the Bullet Cluster and many similar systems), where the gravitational lensing indicates the dark matter has become displaced exactly as theory predicts it should be. Here we see it in in the case of individual galaxies, and we see something similar elsewhere too.

But if instead dark matter isn't a thing, then this is quite the challenge to explain. If it's something else, something related to the visible matter that causes the strange motions, then if you have two otherwise identical systems, then their stars should always have identical motions. There are exceptions. Notably, the external field effect in MOND says that if one system is near another, more massive system, the external gravity can restore Newton-like dynamics in the smaller system. This is why globular clusters, which are well-known in our own galaxy and don't have any hint of dark matter, aren't immediately fatal to the whole idea. 

This option is, however, exceedingly strange and arguably non-physical, but as there's a dedicated recent paper all about the philosophy of MOND, I'll leave that for a forthcoming post. It does seem awfully convenient that we haven't already found more such systems; one would expect this to be a predictable, testable effect, yet the number of systems found which either lack dark matter (or equivalently experience the EFE) is vanishingly small. And it just can't apply to isolated systems, like the UDGs mentioned earlier.

But even if galaxies like this one actually support the dark matter paradigm rather than undermining it, they may still very plausibly challenge the Standard Model in other ways. Nobody expected the existence of large numbers of dark matter-free galaxies, especially not really massive ones. The authors mention that the rotation curve of this galaxy resembles those seen in the earlier universe which have been claimed to be declining at the outer edges, which would also hint at a lack of dark matter. That's not necessarily such a problem for cosmology because galaxy formation is a complex process, though naively one would expect that it's dark matter that drives the accumulation of baryons, not the other way around. That said, I found those claims for the distant galaxies unconvincing. It would be interesting, though, if dark matter was a sort of effect that gradually manifested itself as time progresses, much as dark energy seems to do...

But that's a much more exotic, hand-waving speculation. Best to avoid that. More pragmatically, while dark matter is mass-dominant over normal matter on large scales, locally this is often not true at all.

Unfortunately one the key points of the paper is lost on me. They say that whether the galaxy is compatible or not with the Standard Model depends on where exactly one measures the dark matter content : should it be at the half-light radius (interior to which the stars emit half the light, a standard, objective measure of galaxy size) or at a fixed physical distance value ? And I have no idea what the difference is supposed to be here, since they boil down to doing exactly and quantitatively the same thing, measuring the dark matter enclosed at five times the half light radii or within 6 kpc - which are the same ! And yet they say that if the former measurement is used, this contradicts the standard model expectations, whereas with the latter, the tension is much smaller. How this works, I know not. I'm missing something.

As I said, this paper has some parts which are badly expressed. Though I think it's very wise of them to explicitly refrain from speculations about whether the result really does challenge the Standard Model or not.

Thankfully, their speculations about how such a galaxy could form are on generally safer ground (and easier to follow). Since this galaxy is already very rare as a relic galaxy, it's natural to assume their might be a common cause between these two weird aspects (albeit that this is dealing with VERY small samples). And there's a plausible physical connection here too : dynamical friction. The presence of dark matter tends to act as a gravitational "drag" on anything orbiting within it, making the mergers of satellite galaxies that much easier. As this galaxy doesn't have a dark matter halo, it makes sense that it never experienced the merging phase that would be normally responsible for turning the nugget into a normal galaxy.

So, dark matter-free galaxy => relic nugget instead of a normal galaxy. Good.

But... why doesn't it have a dark halo to begin with ? It certainly seems counter-intuitive, as they say, that such a dense galaxy would form without a correspondingly dense halo - though my own experience suggests otherwise. I saw in my simulations that during a monolithic collapse, the dark matter actually helps to smooth everything out, and without it you get runaway densities.

Anyway, even if the lack of a halo helps explain why this galaxy is a relic, that still doesn't explain why it doesn't have a dark halo in the first place. The authors suggest two possibilities (they say three, but I count two). It could have had a "strong interaction with the environment", with simulations suggesting that dark matter is preferentially stripped because it orbits at generally higher distances from the galactic centre than the stars and gas. Similarly, it could have experienced the same sort of collisional origin that seems to explain some of the other, much smaller dark matter-free galaxies. 

The main caveats to these scenarios is that it's not known if they could explain the frequency of such galaxies, especially since this is the only known massive galaxy without dark matter : if it's due to interactions within the assembly of a galaxy cluster, why don't we see such objects in every cluster ? It also seems doubtful, they say, that tidally stripping the dark matter could account for the very high density of NGC 1277. On the other hand, the cluster environment does seem like a good place for such objects to survive, since the very high velocities of the galaxies means mergers (which would transform the nugget into a normal galaxy) will be rare.

The other option is that it does have a dark halo, it's just much less concentrated than in typical galaxies - so we need observations to greater radial distances to see its effects. Powerful feedback early on, from star formation and/or the supermassive black hole, could in principle expel large amounts of normal matter from the inner regions, with the gravitational field thus being able to disperse the dark matter as well; again, dark matter dominates globally, not necessarily locally. The problem is that feedback this powerful ought to have shut down star formation before it produced such a dense inner core.

So more careful analysis of simulations would help understand if objects like this are expected to be the extreme rarity that a naïve, intuitive view would suggest, or if actually they're not quite such a total anomaly after all (though they will almost certainly turn out to be highly unusual). Better observations, in particular to measure the velocities out to greater distances, would also help constrain the nature of the object, especially to see if it's truly lacking dark matter or only deficient in its innermost regions.

This latter would also help test for the effects of MOND. Unfortunately, this is likely to be extremely difficult. To get to the distance where the accelerations ought to be able to distinguish between MOND and Newton requires distances of 13 kpc, more than twice what they achieve here. And that starts to blend into NGC 1278. Additionally, the very crowded field here surely means the EFE cannot be neglected.

My guess is we'd need to a different approach here : a statistical study of where such galaxies are found and how strong the EFE is. And given how long it's taken to obtain this quality data even for a nice bright NGC object, that's not going to happen any time soon. It's also unfortunate - and I think a wee bit suspicious - that this galaxy looks visually so damn similar to other galaxies, almost as though the dark matter was unimportant...*

* But see that last link. My Master's project was partially motivated by trying to show that dark matter played little apparent role in galaxies apart from making them spin faster; as it turned out, this is just not the case at all. Appearances can be deceiving.

What's the take-home message here ? Well, the good news is that this does appear to be a relic galaxy, which is pretty neat, and it's in a cluster, which is a good place for such objects to survive. Furthermore its lack of dark matter could help explain why it's a relic. The even better news is that we have no idea why it's lacking dark matter to begin with. As with many such anomalies, the fun part is that you can clearly point to them and say, "That thing THERE ! That's WEIRD, that is !". The frustrating part is that going beyond that, following up to confirm exactly how weird it is, whether it presents a challenge to physics or is just one of those things, isn't necessarily any easier just because you know where to look.

More research is needed.

Bing versus Me

A couple of months ago I investigated whether the ChatPDF service could be useful in summarising, analysing and generally discussing astronomy papers. Today I repeat that study, using the same five papers as before and asking basically the same questions, but this time to Bing Chat. Bing uses GPT4 rather than GPT3, so should be a significant improvement in quality. It also has the capability to examine any web page you're currently visiting, making it quite the handy tool provided you don't mind Microsoft's Edge browser. You might remember that I already experienced some mixed but promising results with Bing when using it for more unconstrained astronomy discusssions.

Spoiler : Bing is far better than ChatPDF. It's not perfect, but whereas ChatPDF had major problems with 4 out of 5 papers tested, Bing had major problems with only 1. Of the other 4, it had only minor issues at most, and in at least one case gave an insightful conclusion that I hadn't thought of. Let's go through the papers once again, then, comparing to what ChatPDF came up with as well.

Initially I considered also trying Google's Bard for good measure. However, its results for paper 1 were utterly useless, starting with a description of a totally different paper, and then laced with fabrications, misreadings, statements which were just plain wrong, and horrible inconsistencies. I therefore decided that assessing Bard would currently be a complete waste of my time.

1) Smooth HI Low Column Density Outskirts in Nearby Galaxies

This one I blogged here with the paper itself here. It's about stacking the radial HI profiles of galaxies to see if the column density has an expected break (based on photoionisation models) or if it actually continues to lower values. The answer is the latter.

Unlike ChatPDF, Bing did not decide to inexplicably speak to me in French for no reason. In fact it didn't speak to me in French at all. Its first summary was a good one, but a bit long, but asking it to be shorter retained the crucial points :

The current web page context is a research paper that studies the low column density gas at the outskirts of galaxies as traced by the 21 cm hydrogen line emission (H I). The study aims to investigate the prevalence of breaks in galactic disks and to characterize what determines the potential edge of the H I disks. The authors studied the azimuthally averaged H I column density profiles of 17 nearby galaxies from the H I Nearby Galaxy Survey and supplemented in two cases with published Hydrogen Accretion in LOcal GAlaxieS data. They found that none show evidence of a sudden change in the slope of the H I radial profiles: the alleged signature of ionization by the extragalactic background.

This is ten times better than ChatPDF, which had the highly misleading statement, "The researchers found that this gas is distributed in a more uniform way than previously thought" and gave its unnecessarily evaluative opinion of the importance of the paper. Bing's public outreach summary, however, was lame - far too simple and dumbed down to interest anyone.

At this point I encountered a major problem, not with the AI but with Bing itself. It kept complaining that the current "web page context" was empty, meaning that it had lost access to the content. This led to a truly exasperating (and protracted) back-and-forth which honestly felt like dealing with a bot from ten years ago. Bing asked me if I could provide the web page context, I asked how do I do that, Bing asked me what my question meant, apparently oblivious to our whole previous exchange. 

Good grief. 

It got worse, because even when I figured out it simply meant "open the page in the browser", it would initially answer one question, and then again insist the context was incorrect. It was like talking to the ungodly love child of "whack a mole" and "getting blood from a stone".

Eventually I told Bing to fuck off. This matters, because if a task is faster but more tedious with some new method, I'm going for the slower but less tedious approach nine times out of ten.

I restarted the browser and tried again, this time asking about the lowest density values, the same test I tried with ChatPDF previously. Initially :

Sure! Figure 2 in the current web page context shows the azimuthally averaged HI column density profiles of 17 nearby galaxies. The lowest density values in figure 2 are around 0.1 cm^-2.

Right figure, totally wrong values. Questioning how it got this and whether it was confident led to another back and forth, with yet more "context empty" messages, eventually leading to :

...The paper does not mention which galaxy has the edge of the H I disk similar to what was found for NGC 3198 and M33. Would you like me to look up more information on this topic?

Which was a bizarre and unnecessary statement that doesn't relate to my questions or this paper at all.

I tried again in "creative" mode. And now... things turned a corner. Its public outreach attempt was more than passable :

Galaxies are made of stars, gas, and dark matter. The gas is mostly hydrogen, the simplest and most abundant element in the universe. Hydrogen gas can be detected by radio telescopes, which can reveal its amount and movement in galaxies. Scientists are interested in studying the hydrogen gas at the edges of galaxies, because it can tell us how galaxies grow and evolve over time. Some scientists think that the hydrogen gas at the edges of galaxies can be stripped away by powerful radiation from other sources in space, leaving behind a sharp boundary. However, a new study using radio observations of 17 nearby galaxies found that most of them do not have such a boundary, but rather a smooth decrease of hydrogen gas with distance. This means that other processes, such as gas falling into galaxies or galaxies colliding with each other, may be more important in shaping the outer regions of galaxies. This study helps us understand how galaxies change over time and how they interact with their surroundings.

That point about galaxies accreting gas is arguable but entirely legitimate. Then I asked it for the densities once again :

That is a challenging question indeed. The authors do not state the gas density values directly in the paper, but they do provide the column density values in Figure 2. Column density is the amount of gas along a line of sight per unit area, while gas density is the amount of gas per unit volume. To estimate the gas density from the column density, we need to know the thickness of the gas layer along the line of sight. This is not given in the paper, so we have to make some assumptions or use some external information.

It went on to give an elaborate description of how to convert column to volume densities (this included things like the ideal gas equation), which so far as I could tell was... correct. Its final value was also, broadly, correct. Though not directly connected to the values in the paper at all, I still found this impressive (I only don't say, "very impressive" because I'm not going to rigorously check the working). So I told it I wanted column densities from the paper and it came back with :

I'm sorry, I misunderstood your question. I thought you wanted the gas density, not the column density. The column density values are given in Figure 2 of the paper, as well as in Table 2. The lowest column density value reported in the paper is about $10^{18}$ cm$^{-2}$ for NGC 2976. Is that what you were looking for?

Not bad. The value is basically correct, but it should be for NGC 7331, not 2976. More worryingly, as I'd experienced before, it now proceeded to invention, claiming values for the column densities were given in table 2 - which does not in fact exist.

Summary : ChatPDF gave a poor summary of the paper with misleading claims. It had one flash of brilliance which seems to have been a pure fluke, otherwise tending to get the answers wrong. Its public outreach effort was lame and with results which were qualitatively and quantitatively wrong. Bing gave a good summary, an excellent public outreach statement, and came up with a good methodology for converting the esoteric values of the paper to something more applicable to the real world. However, its claims about specific values still needed vetting, and in "balanced" mode it was a complete moron - though seemingly because of Bing itself more than the language model.

Winner : Bing. Not perfect, but good. I hereafter kept Bing in creative mode.

2) Discovery of an isolated dark dwarf galaxy in the nearby universe

Blog post here, paper here. A much shorter paper that might be easier to handle. This one's about the discovery of an optically dark HI cloud that looks like it's rotating like an ordinary galaxy. The key point is this is isolated, so unlike other such dark galaxy candidates, it's very hard to attribute it to being the result of any other formation mechanism : a primordial origin seems the most likely.

Whereas ChatPDF stated the importance of the isolation in a very strange way, Bing's summary was flawless. Likewise, while ChatPDF had problems finding the very basic quantity of the distance to the object, Bing got it right first time. When asked about why this object was important, Bing have a very good but very generic description of why dark galaxies are important. So I asked it what was so important about this one in particular :

I think the most important characteristic of the cloud is its isolation from other galaxies. This makes it unlikely that it is a tidal debris or a satellite of a larger galaxy. It also implies that it has not been affected by external factors such as mergers, interactions, or feedback that could trigger star formation or gas loss. It is a rare opportunity to study a dark galaxy in its pristine state and learn about its origin and evolution.

Spot-on perfect. ChatPDF noted the caveat that the rotation curve of the object couldn't be guaranteed to be flat, which I agree with. Bing didn't raise this objection on its own, but when prompted :

The paper shows a position-velocity diagram of this galaxy in Figure 1c, which displays an S-like structure that is typical of disk galaxies. The paper also uses a tilted-ring fitting code to estimate the rotation velocity (Vrot) and velocity dispersion (σv) of this galaxy. The paper does not explicitly show a rotation curve plot, but it implies that this galaxy has a flat rotation curve as expected for a rotating galaxy with dark matter. The paper states: “Using the fitted Vrot and σv, we calculated that Mdyn is (5.1±2.8)×109M, which is 47±27 times its baryonic masses, implying that FAST J0139+4328 is dominated by dark matter within the error range.” This means that the orbital speed of the gas does not decline with distance from the centre, but stays constant or increases slightly.

Therefore, based on both web search results and the current web page context, the answer to your question is yes: the rotation curve of the object described in the paper seems like a flat curve as expected for a rotating galaxy.

Which is fine apart from the weird symbols; it's certainly an arguable point whether you'd consider this result solid evidence for a flat rotation curve or not. Pushing back a little on whether this result was secure :

That’s a good question. You are right that the resolution of the FAST data is not very high, and that may affect the accuracy of the rotation curve estimation. The paper acknowledges this limitation and states: “Here, due to the low resolution of the FAST, we can only roughly estimate the content of dark matter. In the near future, we will apply for the higher-angle resolution observation for FAST J0139+4328.” However, the paper also claims that their model can construct the dynamic structure of FAST J0139+4328, except for the weak flat components, and that their results are consistent with the baryonic Tully-Fisher relation for gas-dominated disk galaxies. Therefore, they seem confident that this galaxy has a flat rotation curve as expected for a rotating galaxy with dark matter. Of course, more observations and analysis are needed to confirm this hypothesis and rule out other possible explanations.

Which cannot be improved upon.

Summary : ChatPDF gave inconsistent results about the main finding of the paper, sometimes getting it right and sometimes wrong (or at least phrased extremely strangely). It gave a correct caveat about the uncertainty of one of the main results. Bing unambiguously identified the major distinguishing feature of this object from other, similar structures, and though it didn't volunteer the same caveat as ChatGPT did, it gave a much more detailed, entirely accurate explanation when prompted.

Winner : Bing, due to its better, more consistent answers on the main results.

3) Young, Blue, and Isolated Stellar Systems in the Virgo Cluster. II. A New Class of Stellar System

Blog here, paper here. This is a long one about unusual stellar systems in the Virgo cluster, which the authors think are most likely produced by star formation in gas removed from galaxies by ram pressure stripping.

ChatPDF got confused about the main point of the paper, deciding that these objects could be galaxies in an early stage of formation - which is sort of arguable, but highly misleading. Bing had no such problems. Its summary of the conclusions was more than adequate (the first time it gave a bullet-point summary, with each point containing accurate and pertinent numerical facts, while the second time it gave a paragraph with a perfect paraphrased description of the major findings). When asked why the authors favour ram pressure over a tidal origin, it initially responded :

• Ram pressure can explain the high metallicity of the BCs, which is inconsistent with tidal stripping from low-mass galaxies.
• Ram pressure can explain the presence of H I tails in BC3 and SECCO 1, which are not expected from tidal stripping.
• Ram pressure can explain the isolation of BC4 and BC5, which are too far from any plausible parent galaxy for tidal stripping.

This is decent but imperfect. The first point is a bit oddly phrased, but in another query session, it explained that ram pressure can remove gas from the inner, metal-rich parts of galaxies, whereas tidal stripping preferentially removes material from the outer, metal-poor regions. So that's all good. The second point is more questionable; I'm not sure SECCO 1 has an HI tail, and while the structure of the tails might be different if produced by tides rather than ram pressure, there's no reason to think they couldn't exist. The third point is fine.

On another attempt the second point wasn't mentioned, so I asked it if the objects have HI tails. It correctly answered that only one of the objects did so, naming it correctly. and explaining that some might have emission below the detection limit and/or confused with the Milky Way. That's a point I myself explained when I blogged the paper !

When I pushed a bit more, Bing did unfortunately claim that the galaxies are unusually gas poor. That's not the case at all; as I noted when blogging this, the detection limits aren't good enough to establish this : they could in principle be extremely gas-rich given the current sensitivity of the radio data. And Bing invented a quote and a figure to try and back up its claims. I corrected it, but it then said it was quoting from another paper by mistake, giving a reference... which was to this paper ! I corrected that, so it changed the reference by incrementing the page number in the reference by one. Not good.

In another session I continued down this rabbit hole. Bing was quite insistent that the MHI/L ratio (gas fraction) of both these and other low-mass systems was extremely low compared to other dwarf galaxies, and that even if the upper limit on the HI did allow still a high MHI/L ratio, then it's still likely the objects had far less gas than this. Its response about the nature of the upper limits of HI detections was honestly a little bit incoherent : technically correct if you think it through, but awkwardly phrased and took some disentangling. It also insisted that most Local Group dwarf galaxies have HI detections. 

Here the debate actually became quite interesting. As we found out, in terms of raw numbers, Bing was wrong, but if you slice the data according to isolation (which is actually what you want to do for a fair comparison), it was right. Bing had done the right thing for the right reasons, but only failed to explain its working properly.

Finally I asked it if they could be galaxies in an early stage of formation. It correctly answered no, describing how they're more likely to be transient objects than primordial galaxies. This was perfect.

Summary : ChatPDF fabricated the main result of the paper and invented a quote to support its claim. Bing did this but to a much lesser degree and on more minor points. When pressed, it gave a pertinent discussion revealing its underlying "reasoning" (if we can call it that) was correct, but it just needed to explain it better from the start. It pushed back when I contradicted it, but was willing to take on new information. It made one wholly erroneous claim, but this was not repeated.

Winner : Bing. A bit confusing, but Bing was basically right nearly all of the time, only erring on minor points.

4) The Turn-Down of the Baryonic Tully-Fisher Relation and Changing Baryon Fractions at Low Galaxy Masses

Blog here, paper here. This is an especially complicated and technical one. Low mass galaxies appear to rotate more slowly than the BTFR predicts, but when you account for various corrections, they actually rotate more quickly than the linear version found for higher masses and this can be shown to be consistent with the dark matter paradigm.

ChatPDF did a good summary of the major findings but was at best vague about whether they contradicted the CDM paradigm, whereas both Bing and the authors explicitly claimed that they supported it. Bing's summary was perfect. Its public outreach version was good, but simplified things just a nudge too much. It said that the "turn down" means that very low mass galaxies have slower rotation speeds than expected. Well, they do, but the turn-down itself actually means the relation changes so that we'd expect to see the opposite. However, its explanation for this was actually pretty good :

I was trying to simplify the explanation for a general audience, but you have pointed out a more accurate description of the turn-down in the BTFR. The dark matter profiles of low-mass galaxies are expected to be less dense and more flattened in the center than the profiles of more massive galaxies. This means that the rotation velocity does not reach a constant value at large radii, but keeps increasing. When this effect is taken into account, the BTFR shows a steeper slope at low masses than at high masses. This implies that low-mass galaxies have lower baryon fractions than more massive galaxies.

I asked it if the BTFR contradicted CDM, and it came up with a good description of the general arguments for and against. It didn't take the results of this paper into account though, so I had to be more explicit that I wanted it to do this. And it said that they don't, providing a lengthier and entirely correct explanation. 

Summary : ChatPDF did a good job with the technical details, but erred on one significant point by directly contradicting one the paper's most explicit and important claims. Bing got everything right, though its statements did sometimes need a little unpacking, especially when it tried to simplify things for a general audience.

Winner : Bing, which made no major errors here at all.

5) Atomic Gas Dominates the Baryonic Mass of Star-forming Galaxies at z ≈ 1.3

Blog here, paper here. This much shorter papers stacks lots of HI spectra of galaxies to get a detection of atomic gas at z=1.4, comparing the result with the molecular gas. They find that at redshifts where star formation peaks, atomic gas is the dominant component, which is surprising because it's thought that molecular gas is what's mainly involved in star formation.

ChatPDF had no problems with this one apart from garbling the symbols, and Bing performed equally well in disseminating the paper's content. Both its technical and public summaries were very good. I decided to press it a bit more on the major thing that I found interesting. Namely, that the atomic gas fraction in galaxies increases sharply at greater distances, while the molecular gas fraction doesn't change much. What's odd about this is that it's molecular gas we think is involved in star formation (at least more so than the atomic) and the star formation rate at greater distances is much higher than in the nearby Universe. After a little back and forth, Bing came up with :

Me : I would think that although not at their peak, the typical SFR at z=1 are much higher than today. If molecular gas is the component which dominates SF, I would naively expect galaxies at z=1 to have a higher amount of molecular gas.

Bing : I understand your reasoning, but the current paper's results show that the molecular gas mass of star-forming galaxies at z=1 is not significantly higher than that at z=0. This implies that other factors, such as the gas density, pressure, or temperature, may play a more important role in determining the star formation efficiency than the molecular gas mass alone. In this context, a human might say that the molecular gas mass is a necessary but not sufficient condition for star formation.

Brilliant ! Bing realised that there could be other factors at work, that the simple molecular gas fraction was not the only thing we should consider. This is far superior to ChatPDF perfectly decent but very hand-wavy answer : Bing has offered an actual explanation, whereas ChatPDF only really provided caveats. So impressed was I by this that I went back for another session. A similar line of questioning led to a similar result, even elaborating that the molecular gas must be more efficiently converted into stars and replenished more rapidly from the greater atomic reservoirs at higher redshifts.

Conclusions

I have to say that I'm more impressed by Bing than I was expecting. Do not use it in "balanced" mode, that's a pile of crap. But "creative" mode is showing some serious potential (I didn't yet try "precise"). It's still far from perfect, and it still makes mistakes, and even still just makes shit up, but it does so at a significantly lower rate than ChatPDF. 

You still can't use this for analysis. That is, you can't yet ask it a more general question and trust it will fill in all the blanks for you. For example I asked it the lookback time for a given redshift and it quoted the cosmology calculator, coming back with a number that was well in the ballpark but not exactly right. Neither its raw numbers nor its methodology are yet trustworthy enough to let it run free on its own; see the example of the MHI/L ratios above*. In principle it would be a simple enough matter to compare these values for different samples, but Bing isn't there yet. However, it is pretty good for provoking inspiration of ideas that might genuinely work, and that's... pretty cool, honestly. It can help with analysis, even if it can't actually do any.

* Though I did not encounter anything at all like a previous case where Bing insisted the M/L ratios would always be constant because the units would change, which made not a lick of sense. Possibly this was because I didn't disable the search function, so this gives Bing something to ground itself in.

And... I get the distinct impression that Bing is an awful lot closer to being a genuine analysis tool than ChatGPT/PDF. As an expert, I think you can already benefit from discussions with Bing on academic papers. Bing is good enough to find the most relevant points, and experts know enough to spot any glaring errors; it can also suggest interesting alternative explanations and methodologies. So for pure discussion purposes, I think this is of immediate benefit. I'm almost tempted to say that journalists could use it to help with outreach, but more realistically, they could probably use it at least to find papers that might be interesting to interview the authors about - I don't think it's reliable enough to replace the human component on that front. For now.

The Return of the Failures

Another day, another paper on how exciting Ultra Diffuse Galaxies are. 

At first, these large, faint galaxies were just wholly unexpected, and that was weird enough by itself. But the natural question was how massive they were. Could they be as massive as the Milky Way, having loads of dark matter but hardly any stars for some reason ? Or could they be relative lightweights, more like dwarf galaxies but much more extended ?

Answers have varied over time, but it seems like the picture is at least beginning to clear. Ultra Diffuse Galaxies do seem to fit some scaling relations of typical dwarf galaxies, but it now looks like their dynamics are markedly different. Rather than being "failed" Milky Way galaxies, they're equally surprising but in the exact opposite sense : they have much less dark matter than expected rather than more, in extreme cases even consistent with having none at all. See that last link for a more detailed overview.

Today's paper is another twist in the saga. It says, "hold on there just a minute. Actually some of them might be really massive after all !".

In principle, both of these outcomes are excitingly weird. Nobody ever predicted "failed" Milky Way galaxies, or galaxies having strongly varying dark matter fractions - especially those found in isolation, where we can't just claim it was removed by tidal interactions and suchlike.

I do have some minor quibbles with this paper. I don't remember there being all that much "enthusiasm" for the failed Milky Way hypothesis : sure, everyone would like this to be true because it would punch a massive hole in the standard model, but it always seemed to me that this claim was being made pretty much by one group and nobody else. I also have some issue with saying that the large dwarf hypothesis is a natural "prediction" of CDM; it may well be explicable in CDM, but nobody predicted this population prior to its discovery. Lastly, I don't like the way they keep insisting this work is the first kinematic evidence for very massive UDGs. Yes, the famous Dragonfly 44 is no longer thought to be so massive, but it's stretching the language a bit to then say that this paper is therefore now the first such evidence. Just make the claims, they're exciting enough without the forced superlatives.

That said, I have no issue with the claims at all. I think it's a very detailed, compelling work, and though it unavoidably has issues with small number statistics, I think the authors have done as good a job as possible in trying to mitigate this.

The problem in measuring the rotation speeds of UDGs, which you need to get their total mass, is that they're bloody faint. So here they select UDGs in the Virgo Cluster which have plenty of globular clusters. These have much higher surface brightness so they're easier to measure. Nevertheless, the observations took a long time, beginning in 2017 but delayed due to, "unusual events such as an earthquake, the Kilauea volcanic explosion, a snow blackout, and the pandemic." If that's not sufficient excuse then I don't know what is.

The upshot is that about half their sample of galaxies has about the typical expected velocity dispersion (i.e. total mass) given their optical luminosity, but the rest extends to very much higher velocities. The only potentially significant criticism I can think of here is that it's unclear if their comparison sample (of normal galaxies) is also measured using globular clusters, or if that's done through the stars in the main body of the galaxies. However, it's been previously shown that this method works, though I've not checked the cited paper for this.

It also seems unlikely that they could have misidentified enough globular clusters to have got the measurements wrong. They've been quite conservative in their cluster selection, only choosing those close to the galaxy and at similar velocities, so the chance of interlopers is small. Also some of the galaxies are at quite different velocities to the cluster mean, so the chance of interlopers at the same velocities is even lower in those cases. 

Could they just be galaxies which are being disturbed as they fall through the cluster ? In principle this could perturb them from equilibrium and so the equation calculating the dynamical mass would be invalid. But weirdly, there's a neat anti-correlation between density of the environment and velocity dispersion : the galaxies with the highest dispersions have the fewest neighbours ! If their high dispersions were a result of some cluster-based process, you'd expect to see this more strongly in the densest parts of the cluster. 

I'm putting this one firmly into the "intriguing" category. As they say, no cosmological simulations predict the existence of failed giant galaxies. 

What does this mean for our theory of galaxy formation, then ? At the moment, nothing much. The sample is too small and in a cluster, which are complex places full of weird stuff anyway. It's just not possible to say at the moment if these are "exotica" - weird, anomalous objects, interesting by themselves but not indicative of anything more broadly - or genuinely indicating that something really very strange indeed is going on. We'll see. 

Get Those Mother**ing Satellites Off My Mother**ing Plane

It's been a good long while since I looked at any planes of satellites papers, so let's see what's new in this controversial arena.

For those not in the know, the idea is that there are significant numbers of galaxies whose small satellite companions orbit around them in thin planes. This is not a natural prediction of cosmological models, which show that they should orbit them in spheroidal clouds instead. So this has been seized on as a major challenge to the standard model and even the dark matter paradigm itself.

Now the plane of satellites around our own Milky Way is really very clear and unarguable. But as I go into at great length here, claims for other such planes are not in the least bit convincing, and I tend to view the whole field as awash with some bloody daft statistical biases from people who really should know better. Honestly I think it's just (ahem) plain silly. A very short recent summary that uses a lot of the same arguments that I do can be found here.

But today I turn my attention to this more substantial offering. This paper concentrates only on the plane around the Milky Way. As I've said, this feature is very interesting. I don't think it necessarily means that all of science is wrong, but I'd like to know how it formed all the same.

In this paper the authors use one of the big new all-singing all-dancing cosmological simulations, within which they try and look for Milky Way analogues that have planes. They impose limits on the mass of the host galaxy that are fairly generous, but they have a rather strict policy for considering the 14 brightest satellites of the galaxy. Together with the mass of the satellites, this ensures a direct like-for-like comparison with the observations. At this mass range the observations should be complete (meaning they haven't missed any) and the observations have sufficient resolution to simulate them accurately.

Actually, while their mass range for the host galaxy is quite large, I suspect their constraints on the satellites are if anything too strict. 14 is a pretty arbitrary number, really, and they only report on one single Milky Way analogue that has a plane in agreement with the observations. This bit is my only gripe with the paper : they state their selection criteria for the host galaxy very clearly, but they don't actually state their criteria for a match regarding the plane; my guess is that if they relaxed these parameters a bit, they might find considerably more planes. There is no need, after all, to insist on a perfect match of all parameters.

Anyway, of the 548 halos of the correct mass in their simulations, 404 are as isolated as the Milky Way, and of those 231 have the correct number of satellites. Only 1 of these has a compatible plane (the above caveats notwithstanding), but 1 in 200 is already quite a bit higher than some other estimates. Remarkably, this galaxy also has a very similar large-scale environment to the Milky Way, with its nearest massive neighbours matching the properties of Andromeda and Centaurus A pretty well. These environmental conditions weren't used as a selection criteria at all.

The plane itself is also very similar to that observed. Not only does it have a similar geometry but it's also rotating, and at a very high inclination angle with respect to the stellar disc of the host galaxy. This is really an excellent match; even if the mass of the Milky Way analogue is a bit low, it's still within observational constraints on the Milky Way. To match this closely on so many parameters at a rate of 1 in 200 is very impressive result indeed.

Now one of the frustrating aspects of many anti-standard-model papers is that they search simulations for analogues of the observed planes, find some, and then only compute the frequency at which they're found and nothing else. Observational planes, they say, are very common, whereas simulated planes are very rare. They conclude that this means there's a conflict between the standard model and observations. The problem here -  well, one of them - is that they seldom if ever attempt to examine how those few planes that they do find actually form in their simulations... and this can make a world of difference. 

What I mean by this is that saying that they're rare overall in the simulations might be correct but potentially irrelevant. For example, giant redwoods are very rare among trees, but if you walk through Giant Sequoia National Park you shouldn't be amazed that if you've found far more than random chance would suggest. The approach of those claiming the planes challenge the standard model is in essence entirely statistical, neglecting the physical processes at work, treating galaxies are random when in fact they're anything but.

Here the authors find that the plane results from two mechanisms. First, the satellites infall along large-scale filaments, preferentially leading to the formation of elongated structures. Second, while most of the satellites here are indeed orbiting around the galaxy, three are just coincidences : their true 3D velocities will take them in quite different directions, so it just so happens that at one particular moment, the plane appears to have more members than it really does. It's partly "real", but partly transient.

At this point I want to raise another issue. Some years back, having a protracted email discussion with a rather strongly pro-plane group, one response about the infall of filaments was rather brief :

"I had already talked to Libeskind, and he never even wanted to suggest that the filaments are related to the VPOS. They are far too thick anyway."

Well, Libeskind is one of the three authors of this paper, and not only do they here explicitly state that the filaments are part of the formation mechanism, but they several times cite his earlier papers as claiming that as well. The above quote is thus demonstrably not correct. You can see why these wholly erroneous claims tend to annoy me quite a lot.

Is this the last word on the matter ? No, though it probably should be. If planes really were as common as some suggest, with practically every nearby massive galaxy having a planar system, we'd likely have a big problem. The authors here are careful to state that their findings don't say anything about these other systems, but in my view, none of these other systems are even remotely comparable to the Milky Way system and really aren't worth bothering with. 

In short, models do predict planes. Not many, but more than some claim, and those which form seem to be in very good agreement with the observations both on large and small scales. By showing that there are physical processes leading to the formation of planes, this is a strong rebuttal of claims that planes pose a serious challenge to the standard model. I remain convinced that this, like many claims against the standard model, is a non-issue.

AI-Assisted Astronomy ?

Yesterday I decided to stop feeding the chatbot weird premises for crossover stories and try and use it for research. Not actual research, you understand, just a test. I did a similar exercise recently to see if a bot could properly fulfil its advertised ability to summarise papers and found it badly wanting, so my expectations were set pretty low. For this one I wanted instead to discuss something explicitly similar to my own research, so I wouldn't have to check all the answers because I pretty much already know them by heart. This was prompted by idle curiosity and learning that Bing AI is powered by GPT4, which is supposed to be a substantial improvement in terms of accuracy.

1) ChatGPT : Up to its old tricks again

First, I had a protracted discussion with ChatGPT about the possible nature of an extragalactic hydrogen cloud. I gave it some basic properties : line width, radius, distance, wavelength of detection. Then I asked it to speculate about its possible origins. Overall, it did pretty well at this, suggesting it might have been stripped from another galaxy or a primordial object that hadn't formed stars. I had to push it just a little to get it to estimate the dynamical mass and realise the key point that the object should be dark matter dominated, but it came up with results which were decently accurate. It came up with a genuinely good list of the other sources of motion in such a cloud besides rotation, noting that these were likely to be small in comparison, hence the need for dark matter.

At bit more prompting and it got a decent estimate for how long such a cloud could survive. It came up with a couple of correct examples of similar known objects too. I moved on to ask it about whether this high dynamical mass could be the result of stars and maybe the SDSS just wasn't sensitive enough to detect them. It produced a decent order-of-magnitude formulae to estimate this, but then it started to break down. It kept giving ever-more inconsistent numbers (I'm actually surprised it made it this far, since this is the free version which doesn't do actual calculations at all). When its mistakes were pointed out, it was pot luck as to whether its revised response would be any better or not. Still, its basic method for estimating the detectability of the stellar content, though crude, is something genuinely useful that I hadn't thought of before. And its list of suggestions for further research to help properly nail-down the cloud was absolutely 100% spot on.

After that it seriously degenerated when I asked it for a summary suitable for an academic paper. It started inventing all kinds of extraneous details, even deciding to give the cloud a plausible-sounding catalogue name, contradicting itself with regards to numbers, and deciding that the cloud had been detected in an optical survey despite explicitly being optically undetected. It even included irrelevant references for some reason. All in all, this part of the test was generally just unhelpful garbage. This was surprising and disappointing, because this is the sort of thing I'd expect ChatGPT to be good at.

In summary, it provided some useful ideas even at the expert level, but its specific numbers were, totally unsurprisingly, not at all reliable, and only when I prompted it did it admit it wasn't doing any calculations - something it absolutely should have been up-front about. I like that it's useful for exploring new ideas, but while this is beneficial, it's hardly revolutionary.

2) Bing Chat : A glimpse of the future or a freakishly coincidental hallucination ?

This one was truly strange, to the extent I almost wonder if I dreamed the whole thing. Bing AI is annoying for two huge reasons (besides, well, being Bing). First, you have to use Edge to run it (FFS, let me choose my own damn browser), and second because it gives you no easy way to save your history. It's either old-school copy+paste or nothing. And since I was on mobile, when I closed the app for a moment, all was lost. This is completely stupid.

At first it didn't look hopeful at all. In "balanced" mode it straight-up refused to give me anything useful in the way of an answer, shutting down the conversation completely so that you can't enter any more text, leaving you with no option but to start over. Why in the world anyone thought that giving it this "ability" was a good idea, I don't know. Again, this is stupid and frustrating, even for a preview tool.

And then... something truly amazing happened. In "precise" mode, it... did exactly what I wanted. True, it needed a little prodding, as ChatGPT did. But it also offered explanations that ChatGPT hadn't considered. It came up with citations as clickable links. It gave the formulae and, impressively, its numbers were absolutely self-consistent. It never messed up the masses of different components as ChatGPT did.

For an estimate of the optical detectability of the cloud it did (or at least appeared to do) something much more sophisticated than ChatGPT. It initially even said this was impossible without running a full population synthesis model with Staburst99, which it can't do and I'm not going to either. Then I told it to make a simpler estimate, and it required the stellar distribution (I told it to assume a standard IMF) and composition (I told it to assume solar metallicity). It then estimated the luminosity, assuming all the massing mass was stellar. It gave a value in Watts (I presume it defaults to SI units, which is not unreasonable) but had no problem converting into the more familiar solar luminosities and then apparent and absolute magnitudes.

I did not have the opportunity to check those numbers, but I do know they were perfectly credible. I'd really have love to scrutinise its calculations minutely, but I was sadly denied this opportunity. But they were certainly close to what I was expecting. I don't know if Bing AI has access to some mathematical tools (like the Wolfram Alpha plugin for paid versions of ChatGPT), but it certainly seemed like it was doing calculations and not just generating numbers statistically.

Bing AI stressed that these numbers were estimates and subject to a lot of uncertainty, something ChatGPT didn't do. I pushed it further, asking if it would be possible to alter the metallicity and/or IMF to render the galaxy undetectable, as with these simple assumptions the galaxy should be well above the SDSS sensitivity limit as I was expecting. It said yes, but when I asked to to check how, for example, the metallicity needed to do this compared to known galaxies, it found that the result was incompatible with known observations and gave me a reference to extreme metallicity values. Similarly for the IMF.

If correct, this is incredibly useful. A lot of tedious calculations and trawling papers... all gone, replaced with a quite natural style of conversation that gets right to the point.

And then all was lost forever. Worse, this morning Bing refuses to do any calculations at all (except in "creative" mode, which produces results which are wrong by many orders of magnitude). It won't give me the stellar mass estimate or even the dynamical mass. It comes up with formulae but its responses are partly garbled as it's very blatantly just scraping together bits of relevant text from different sources (it's at least honest about this and provides the links), and it point-blank refuses to admit it can do calculations at all. It even suggested I may have confused it with another chatbot. And to be fair, the experience is like using a different AI altogether, as though some bloke called Dave crawled around inside it and starting pulling out vital circuitry.

Well, I don't know what to make of all this. ChatGPT did better than I expected, and if that Wolfram Alpha plugin works as advertised... this could be extremely powerful. But as it stands, it's useful for discussions and ideas, but not actual analysis, and somewhat surprisingly, not for constructing replacement text either .

Bing, on the other hand... it might have been total garbage for all I know, that just happened to get things about right. But if (and I do stress "if" very strongly !) this is what using a language-model AI coupled with a genuine mathematical calculator is like, then it's transformative. I want this. Anyone saying it isn't useful is simply mad and wrong.

Dynamically Different Dwarfs Defy Definitions

This paper is another page in the long-running saga of the dynamics of Ultra Diffuse Galaxies. Many of these large but faint objects have been claimed to have a curious lack of dark matter, in contradiction to what we might guess from standard cosmological models. That by itself does not challenge the dark matter paradigm, but might even, perversely, support it. 

How ? Well, the existence of dark matter at least allows for the possibility of this separation of components, whereas theories of modified gravity basically don't. In the CDM (cold dark matter) framework, galaxies may or may not contain dark matter, and this mass strongly affects the motions of their stars and gas. Indeed, we know that not everything is dominated by dark matter : globular clusters and tidal dwarf galaxies certainly don't have any dynamics that suggest any significant amount of extra mass.

In contrast, with modified gravity theories (or Modified Newtonian Dynamics at any rate), if you have two systems with the same distribution of mass, they should always have the same dynamics. These theories say it's all dynamics is due to the visible material, so if this is the same, so should be the motions of the stars and gas. A major caveat is that the systems must be isolated and in equilibrium, but that's the long and short of it.

But the discovery of isolated galaxies apparently lacking dark matter poses a different sort of challenge to the standard model. Such systems weren't predicted; there's no obvious reason to expect them to exist. And yet they do.

This is a convoluted tale, and it's been a while since we last looked at this. So now's as good a time as any to recap the whole story so far. Feel free to skip ahead if you prefer.

---

1) Ultra Diffuse Galaxies were first discovered decades ago, but not as a class of objects. They were noted as exceptional, "low surface brightness" galaxies, big and faint and fluffy, but rare. Then a few years ago they were discovered in large numbers, first in galaxy clusters, but then in other environments, even in isolation. Only when they were realised to be common were they given the UDG label to distinguish them from more general low surface brightness objects.

The obvious question after their discovery was whether, being large, these were also heavily dark matter-dominated systems, "failed" Milky Way-like galaxies only much fainter, or actually a sort of huge dwarf : very extended structures which had some dark matter component but nowhere near as much as the true giants. Although there are a few very interesting objects which may well have a great deal of dark matter, the consensus seems to have settled fairly quickly on "huge dwarfs" rather than "failed giants". I go through this in a lot more detail here and here. 

The most extreme interpretation of this is that UDGs are actually misnamed, and their large size is a misleading result. There are a lot of different ways to measure galaxy sizes, and only one of these really indicates that they're giants. Using other measurements, they're actually of the same size as typical dwarfs. They'd have unusually flat distributions of matter, but this isn't nearly exciting as the more popular claim that they're as big as the Milky Way but a thousand times fainter. Certainly it's a lot easier to account for the discovery of lots of faint dwarf galaxies than it is lots of giant ones. Cosmological models in fact demand we find extra dwarfs, but would suffer horribly if there were lots more giants than predicted. A few would be okay, but a whole population would lead to some serious head-scratching.

So this general picture seems to point to UDGs as being interesting for the details of galaxy evolutionary theory, but not a challenge to anything more fundamental. But then things took a left turn.

2) Measuring the dynamics of these systems is a challenge because they're so bloody faint. Some early results did show that at least some were indeed dark matter dominated, but not to the extent of being giants. But then came the discovery of two UDGs, in a nearby group, that apparently lacked any dark matter whatsoever. Controversy dogged this one from the start, with initial claims that they'd just measured the velocity dispersion wrong (I thought this one was a bit forced, and it turned out the measurement was just fine) giving way to a much more protracted debate about the distance. 

Finally it turned out that the distance was the larger, more interesting value and the claims that they lack dark matter were vindicated. This is an important lesson in not trusting to consistency too much. Had they been smaller and closer, they'd have been perfectly normal systems and it would have been a big hoo-hah about nothing, which is usually the case. But this time it was the more interesting result that was correct.

That places UDGs firmly back in the realm of being a potentially serious difficulty for the standard cosmological model. Yes, we know such objects can form in rare cases, but nobody predicted them in large numbers like these. Indeed, when people did find them in their simulations, they found they were the result of a numerical error, not a physical result at all !

But along came a new model which seemed to be quite satisfactory. It's well-known that tidal encounters between galaxies can preferentially remove dark matter, since this is their most extended component. This new one showed that direct collisions can, in the right conditions, remove enough dark matter to explain these two weird galaxies, in a rare but not exceptional event. Further observations seemed to vindicate this.

Thus was the challenge to cosmology averted once more. Initial worries that there might be too many massive, faint galaxies gave way to concerns that there might be too many faint, dark matter-deficient galaxies, but these were addressed with a convincing explanation : only a few of these latter types had been found, and those in an environment where galaxy-galaxy encounters could be shown to be plausibly responsible. Phew !

All was well with the world. Except...

3) These two cases were the tip of the iceberg. Quite separately, some UDGs had been found to have gas, which makes measuring their dynamics much easier. And a whole population of these appeared to be rotating too slowly, again implying a lack (or at least a deficiency) of dark matter. Even with really good quality data this still seemed to be the case and not a measurement error. And they were in isolation too, so the collisional model just couldn't work for these.

But then it seemed that actually maybe there was a measurement error after all, and that while the authors had put forward many genuinely excellent arguments to support their conclusions, it was more probable that they're just estimated the inclination angle wrong. It only takes a slight error in this angle to turn an interestingly-small rotation speed into a boringly-high rotation speed, to go from no dark matter to an entirely normal amount of dark matter. Though, this is only my interpretation, and the authors are sticking to their guns that they've measured the inclinations correctly, so fair play to them.

There's one other, largely overlooked part of the story : the detection of a population of normal brightness galaxies which rotate too slowly. Here, claims that the inclination angle is a problem don't look credible to me, and these probably deserve more attention.

So that's where we currently stand. The discovery of UDGs initially prompted concerns that maybe there was a population of massive galaxies that could pose a severe problem for cosmological models, maybe even for the dark matter paradigm itself. That concern has greatly alleviated. Now the question is more the opposite. Are there a significant number of isolated galaxies which have far less dark matter than more typical galaxies ? If so, how do they form, why aren't they found in our models, and why didn't we notice them before ? 

---

Enter the latest paper. Previously, analysis of the gas-rich UDGS (HUDGs, for HI-rich UDGs) has concentrated on the half-dozen or so with high-resolution gas measurements. This allows us to check that the velocities of the gas follow the standard, ordered pattern expected for rotation. It also allows us to estimate the inclination angle independently of the optical data, as the HI and stellar discs might be misaligned. We need the inclination as this lets us convert the apparent rotation speed (that we directly measure) into a much more accurate value.

This paper revisits the first sample of HUDGs. I noted at the time that these seemed to show a distinct trend from the Baryonic Tully-Fisher Relation, which simply plots total mass of normal matter as function of rotation speed. Traditionally galaxies follow a very nice, low-scatter trend here, but the UDGs extended to much lower rotation speeds.

My plots were impossibly crude (for a gruelling blow-by-blow account of how to do all the proper corrections needed, see this), and here the authors attempt something much more rigorous. They restrict the sample to the best possible cases, 88 galaxies where the optical images are clear and the HI signal is bright. They don't have high resolution gas data, but they have a clever way to demonstrate that this isn't necessary.

Their main result is exactly as I found : the HUDGs demonstrate a break in the BTFR, following a continuous sequence from "following the usual relation" to "having much lower rotation speeds than expected". Although this is in no small part because they have a great deal more gas than typical galaxies, interestingly they still show the same basic offset even just using the stellar mass rather the total (gas and stars combined). 

Of course the question then becomes, well what about the inclination angle ? Here's their clever trick. They don't have any better gas measurements than the original sample, so these are all unresolved : they can't check for sure if these objects have ordered, rotational motions. So they use the optical measurements to estimate the inclination angle. Using this, when the plot the corrected rotation velocity as a function of the apparent axial ratio of the galaxies (which is what gives you inclination), they find no evidence of any systematic trends. This means that inclination angle errors might be causing some extra uncertainty and scatter in the velocities, but are very unlikely to be responsible for the overall systematic trend.

This looks like a very nice piece of work for me, quite careful to check for any possible systematic effects, plotting things in many different ways to figure out what's going on. The one trick they've missed is a reference to this recent paper, which shows that some low-mass galaxies (not specifically UDGs) which apparently rotate too slowly can be reconciled with the BTFR with enough corrections. So their claim that the deviation of the UDGs from the BTFR is not in agreement with standard models might not be correct, but on the other hand, this is all very new and it needs a lot more examination. Furthermore, that other paper was concerned with galaxies which were much lower mass than these, so its finding may not be all that relevant here.

Where does that leave us ? Well, these UDGs seem to be dark matter deficient but not totally dark matter free. They say, "The formation of them cannot be reproduced in current cosmological simulations. These baryon-dominated dwarf galaxies could be game-changer laboratories in testing cosmology models and galaxy formation models." And they might be. They're weird. But this saga is already full of many twists and turns, and this paper is just one more page in what's turning out to be a gripping read. What happens next, I for one wouldn't like to guess.

ChatPDF Versus... Me !

The latest offering from the wacky world of chatbots is something that claims to be actually practical.

Now, I love ChatGPT and simply cannot get my head around claims it's not useful. I'm fairly sure that this is just my ultra-cynical social media feed, but I will, however, cheerfully concede that ChatGPT's factual accuracy is less than impressive, as its its moral judgement. This is something that claims Poland is a landlocked country, won't output a funny version of the climax of Lord of the Rings ("because the events are meant to be taken seriously"), but also has no qualms about writing a bunny-based version of the Holocaust.

Go figure.

Anyway, if you approach ChatGPT as some kind of improved search engine then you're idiot plain and simple. That's just not what it's good for, any more than expecting a paintbrush and an easel are going to revolutionise the world of journalism. It doesn't make any sense.

But ChatPDF ought to be different. This is specifically designed to read and interpret the contents of PDF documents, including figures. It can summarise the main points and answer questions about what's in it. Now this is potentially a very useful tool indeed, as all too often, reading papers is a horrible chore. Sometimes I just need to find a specific value or equation and trawling though the whole thing isn't fun, with the Ctrl+F function being of limited help, especially with mathematical symbols. So I would hope (not expect !) that this would have at least a bit more rigour to it.

For reference, you can give a link to a PDF to ChatGPT itself, but it's shite. I sent it this one (selected at random), which is a paper entitled, "On the Reionization-Era Globular Cluster in Low-Mass Galaxy Eridanus II". It complained to me several times that the link doesn't work (it does, and arXiv's downtime is negligible, so that's just erroneous). When I told it I could see the paper in my browser, it came up with this :

I apologize for any confusion. It is possible that there was a temporary issue with the link earlier. I have tried accessing the link again, and this time it appears to be working properly.

The PDF you provided is titled “Exploring the Potential of Deep Learning for Automatic Music Genre Classification.” The paper discusses the use of deep learning techniques for automatic music genre classification. The authors conduct experiments on a large dataset of music recordings and evaluate the performance of various deep learning models. They compare their results to those obtained by other classification algorithms and find that the deep learning models outperform these other methods. The authors conclude that deep learning shows great potential for improving the accuracy and efficiency of music genre classification.

Which is pure garbage. Perhaps worse, when I gave it the correct name of the paper, it gave me a very plausible-sounding summary. Asking it where it got a value of the radius from, it confidently declared it was from a section that didn’t exist. Further corrections just led to a whack-a-mole of the same thing. This is dire, because it's all too easy to fall for its surprisingly coherent drivel.

But testing a random paper isn't a good idea since I've no idea what's in it. What would be a much better way to test this is... this blog ! More than 200 papers summarised by a source I can't have any higher level of confidence in, because it's me.

For this test I'm going to take the five most recent papers on this blog and use the specialised ChatPDF. Let's see if this does better than the ChatGPT app and how it compares to doing to hard work of actually reading the paper.

1) Smooth HI Low Column Density Outskirts in Nearby Galaxies

This one I blogged here with the paper itself here. It's about stacking the radial HI profiles of galaxies to see if the column density has an expected break (based on photoionisation models) or if it actually continues to lower values. The answer is the latter.

At first ChatPDF decided to speak to me in French. No idea why and it doesn't do it for other papers (the authors aren't French and they don't work in French institutes). So I told it to speak English instead it obliged. I asked for an explanation but all I could get was an apology. Weird.

Anyway, its summary of the paper was decent but not any better than the abstract. When I asked it for the main conclusion :

The main conclusion of the paper is that the outer regions of nearby galaxies contain smooth, low-density gas that is likely to be an important source of fuel for future episodes of star formation. The researchers found that this gas is distributed in a more uniform way than previously thought, which has implications for our understanding of how galaxies form and evolve over time. Overall, the paper provides new insights into the complex interplay between galaxies and their surrounding environments.

Initially I thought this was pretty good but on reflection it's actually very poor. Whether the gas could sustain future star formation is irrelevant : kudos for making the connection, but the authors don't so this is not warranted (also the total mass of the gas is unlikely to be significant). "More uniform" is arguable. They do use the word "smooth", but in context this is referring the 1D azimuthally-averaged profiles, i.e. they don't have any distinct break. That the average density of the gas decreases smoothly in profile doesn't mean the local density of the gas isn't highly variable. And the last sentence and a half are very much in the style of ChatGPT in that they contain no actual information but essentially give an opinion about the quality of the paper. They're okay for a general audience who need to know the broader relevance but of no use at all for a specialist who already understands this.

So far, so mediocre. Next I tried asking a more specific question about the lowest density value obtained. Initially it said this wasn't given, but then I corrected myself to column density. It said :

The paper states that the column densities of the gas in the outer regions of galaxies are low, dropping below 10^19 cm^-2. The researchers found that the gas is distributed smoothly and does not exhibit a sharp break in column density, which is contrary to what some photoionization models predict. This suggests that ionization by extragalactic photons may not be the limiting factor for the extent of the HI disk, and that instead, the outskirts of the disk may correspond to a transition to a low column density gas accreted from the cosmic web at later evolutionary stages of disk formation.

Now this is 100% correct and cannot really be improved upon. It gave a similarly impressive summary of the methodology, referencing the page numbers and giving the right quantitative values. On pressing for an exact value of the lowest column density measured, it reported that the authors don't give an exact value, which indeed they don't. So I tried :

Could you estimate the lowest value reached using their figure 2 ?

Based on Figure 2 of the paper, it appears that the lowest column density reached is around 5 x 10^18 cm^-2. However, this is just an estimate based on the graph and not a precise measurement.

Which is seriously impressive stuff. It would have been better if it had volunteered this itself, but that it can do this at all is amazing.

Except... it's a fluke. Repeating this in a later session it at first insisted the value was stated in the file itself, 10^19, which is confusing a limit with a measured value. Asking it to use the figures instead didn't work. Asking it for figure 2 specifically reverted to 10^19, which is just not right at all.

Finally I asked it for a public outreach summary. The result was pretty poor, no better than I could obtain from rewording the abstract, and when pressed, it insisted that the density values obtained were much higher than in ordinary air : this is wrong but about 25 orders of magnitude.

Oh dear. This isn't looking good at all.

2) Discovery of an isolated dark dwarf galaxy in the nearby universe

Blog post here, paper here. A much shorter paper that might be easier to handle. This one's about the discovery of an optically dark HI cloud that looks like it's rotating like an ordinary galaxy. The key point is this is isolated, so unlike other such dark galaxy candidates, it's very hard to attribute it to being the result of any other formation mechanism : a primordial origin seems the most likely.

The summary of this one was okay, but when I asked it for the main conclusion, it got hung up on the fact that the cloud is isolated. This is important, but it stated things in an almost tautologous way, as though isolated clouds should be expected to have nearby companion galaxies, which is obviously nonsense (otherwise they wouldn't be isolated). However, in a reverse of the first paper, when I asked it again in a new session, it gave a much better answer giving a direct quote for the paper that was absolutely relevant.

More disappointingly, it initially couldn't give the distance to the cloud, saying that the authors didn't state this. At first I thought this was correct and they just hadn't mentioned it, so I asked for the distance based on its systemic velocity. Now it gave the correct value. But later I found that they do actually state this value directly, so this is no more impressive than doing a Ctrl+F for "distance". On the other hand, it correctly reported that the rotation curve can't be guaranteed to be flat, which would be the key signature of a dark matter-dominated system.

So far, so very ChatGPT-like. Occasional flashes of genuine brilliance wrapped in large doses of mediocrity and sprinkled with total rubbish. Let's try another one.

3) Young, Blue, and Isolated Stellar Systems in the Virgo Cluster. II. A New Class of Stellar System

Blog here, paper here. This is a long one about unusual stellar systems in the Virgo cluster, which the authors think are most likely produced by star formation in gas removed from galaxies by ram pressure stripping.

The summary of the paper wasn't great, saying that :

The authors suggest that these systems may represent an early stage in the formation of larger galaxies, or they may be remnants of disrupted galaxies that were once part of a larger group or cluster. 

Which is certainly not the main point at all. When I asked for the most likely origin, it did very much better :

The authors suggest that the most likely origin of these young, blue, and isolated stellar systems in the Virgo Cluster is ram pressure stripping. Ram pressure stripping occurs when gas is removed from a galaxy as it moves through the intracluster medium at high speeds. This process can lead to the formation of new stars in the stripped gas, which may then form into compact, low-mass systems like those observed in this study. The authors propose that these systems formed from ram pressure stripped gas removed from new infalling cluster members and survived in the intracluster medium long enough to become separated from their parent galaxies by hundreds of kiloparsecs. However, they note that tidal stripping cannot be conclusively excluded as a formation mechanism for these objects.

Spot-on perfect. Going back to its other claims, it told me that the possibility of being an early stage of galaxy formation was mentioned on page 2, giving a quote. But this was just flat-out wrong as this quote just doesn't appear anywhere in the paper at all. The same was true about the idea of being disrupted remnants, giving a quote and page reference that was a barefaced lie. Telling it it's made mistakes does have it correct itself, but this really shouldn't be necessary.

 Dear oh dear oh dear.

4) The Turn-Down of the Baryonic Tully-Fisher Relation and Changing Baryon Fractions at Low Galaxy Masses

Blog here, paper here. This is an especially complicated and technical one. Low mass galaxies appear to rotate more slowly than the BTFR predicts, but when you account for various corrections, they actually rotate more quickly than the linear version found for higher masses and this can be shown to be consistent with the dark matter paradigm. ChatPDF's summary is quite correct :

The main conclusion of the paper is that the baryonic Tully-Fisher relation (BTFR) turns down at low galaxy masses, which indicates a decrease in the efficiency of galaxy formation. This turn-down is likely due to a combination of factors, including feedback from star formation and accretion onto dark matter halos. The authors suggest that further study of this phenomenon could help to resolve discrepancies between observed and predicted galaxy mass functions and improve our understanding of galaxy formation physics.

I asked it to describe this in terms of galaxy rotation speeds and was even more impressed :

The turn-down in the baryonic Tully-Fisher relation (BTFR) at low galaxy masses refers to a decrease in the efficiency of galaxy formation, which is reflected in the relationship between a galaxy's baryonic mass and its maximum rotational velocity. Specifically, at low galaxy masses, the observed rotational velocities are lower than what would be expected based on extrapolation from higher mass galaxies or from galaxy formation models. This suggests that there is a decrease in the efficiency of converting baryons into stars and/or a decrease in the amount of dark matter relative to baryons at low masses.

Asking it for the observational comparison with the BTFR from high mass galaxies also gave a perfect comparison. I asked it for the figure illustrating this and it correctly picked figure 5. Asking it if this could be reconciled with the CDM paradigm was more disappointing, as its answer amounted to a cautious "maybe" rather than describing the author's claims. Pointing to figure 9, where the authors do demonstrate how this reconciliation may be possible, it insisted that they didn't. 

This is disappointing. Asking in a somewhat more technical matter gave an answer that was both better and worse :

Describe this in terms of rotation speeds, comparing with the predictions of a linear BTFR.

The right panel of Figure 9 shows that the observed rotational velocities of galaxies with baryonic masses less than 10^8 solar masses are higher than what would be expected based on a linear extrapolation of the baryonic Tully-Fisher relation (BTFR) from more massive galaxies. This indicates a turn-down in the BTFR at low galaxy masses, which suggests a decrease in the efficiency of converting baryons into stars and/or a decrease in the amount of dark matter relative to baryons at low masses. The green line in the figure represents the best-fitting linear relation between maximum rotational velocity (Vmax) and baryonic mass for galaxies with masses greater than 10^8 solar masses. The galaxies with lower masses that lie to the right of this line have higher Vmax values than what would be expected based on this linear BTFR. This discrepancy between observed and predicted rotation speeds at low galaxy masses is a challenge for the standard cold dark matter (CDM) model and may require modifications or additional physics to explain.

The problem is they say even in the abstract that this is consistent with CDM ! Once again I was given an invented quote. Now, this claim is reasonable but misleading. It's reasonable that we might not fully understand the physics of galaxy formation, especially at low masses, but the point of the paper is to demonstrate that this change of slope in the BTFR is consistent with CDM predictions. Saying it requires additional physics gives a completely different impression of the claim being made.

5) Atomic Gas Dominates the Baryonic Mass of Star-forming Galaxies at z ≈ 1.3

Blog here, paper here. This much shorter papers stacks lots of HI spectra of galaxies to get a detection of atomic gas at z=1.4, comparing the result with the molecular gas. They find that at redshifts where star formation peaks, atomic gas was the dominant component, which is surprising because it's thought that molecular gas is what's mainly involved in star formation.

ChatPDF's summary does not begin well :

We found that the baryonic composition of star-forming galaxies at z≠1 is dramatically different from that of galaxies at z≈0. 

At redshifts not equal to one, eh ? Okay then. It also garbled the symbols in the rest of its responses. However, its statements were generally on point. It completely "got" the central point that the greater atomic fraction during the peak of star formation is surprising, e.g. :

Our study did not directly investigate which type of gas, molecular or atomic, is more important for star formation. However, we did find that the average atomic gas mass of star-forming galaxies with 〈Må〉≈10^10 Me is comparable to the average stellar mass at z≈1 and is significantly larger than both the average stellar mass and the average molecular gas mass at z≈1.3. This suggests that atomic gas may play an important role in star formation in galaxies at high redshifts. 

That being said, it is generally believed that molecular gas is the primary fuel for star formation in galaxies (e.g., Kennicutt & Evans 2012). This is because molecular gas has a higher density and lower temperature than atomic gas, making it more conducive to forming stars. However, atomic gas can also be converted into molecular gas through various physical processes such as cooling and compression (e.g., Krumholz et al. 2009), so it is possible that both types of gas are important for star formation in different ways and under different conditions.

This is pretty perfect, references (which I'm not going to check) and symbols notwithstanding. I asked it for how many redshifts they considered and it correctly differentiated between the two new data points from the study plus the comparison sample at redshift zero. For this paper at least, the only fault seems to be the peculiar symbols. Its factual accuracy was in this case fine.

Conclusions

I'm afraid this one can't be said to be anything beyond the usual "impressive tech demo" stage. It is categorically not ready for actual use and anyone paying the subscription fee is at this stage a complete fool. 

To give credit where credit is due, it does often produce remarkably good summaries that are more accessible than reading the abstracts. It can extract complex variables, even ones which are stated directly in the text. It seems to do better when you ask it very specific questions, but it's capable of handing complicated technical descriptions and distilling them down to their most relevant points in even more generalised ways.

The problem is that accuracy and usefulness do not scale linearly with each other. If it produces accurate statements 70, 80, even 90% of the time, it's only useful 0% of the time. Why ? Because that failure rate is such that its claims always have to be checked, and one would be better off just reading the paper. You have no idea if it's just making stuff up or missing a vital point. Worse, it's dangerously coherent. If you're not already an expert in the field, it produces statements which sound fully convincing but are in fact just plain wrong. I'm glad it references the parts of the text it's getting its information from, but it frequently just invents entire quotes, and that's unacceptable. 

Garbage in garbage out ? In this case it's sometimes "gold in, garbage out", which is worse. It has the frustrating tendency to veer wildly between a superb, concise description and stuff which is pure fabrication, or underperforms compared to just doing a regular Ctrl+F.

That said, thresholds are important. Were it to reach, say, 95%, 99% reliability, then usefulness might well rise markedly and steeply above 0%. It doesn't need to be 100%, because reading papers doesn't produce a 100% accuracy either. It just needs to do a lot better than its current level. I'll be keeping my eye on this one, but, dear reader, it looks like this blog will fulfil a valuable purpose for the foreseeable future.