80 Comments
User's avatar
Boltzmann Booty's avatar

My guess: atmospheric electrical discharges that become more common in the wake of nuclear explosions due to perturbed ionization or something. Would explain both the correlation with nukes & the correlation with UFOs (conditions that produce flashes above telescope plausible would also tend to produce odd aerial phenomena visible to people). Thinking like ELFs, sprites, etc. But don’t know if that somehow is invalidated by the statistics of appearance in light vs shadow (though could also imagine day/night affecting atmospheric electrical phenomena, which could impart a bias on flashing)

Expand full comment
Robin Hanson's avatar

Stuff in the atmosphere is ruled out by the shadow analysis.

Expand full comment
TGGP's avatar

My reading of the shadow analysis was that it only excluded plate defects.

Expand full comment
Robin Hanson's avatar

The shadow analysis says most of the effect is in stuff that is changed by being in shadow. The telescope only looks up at night, so anything close to Earth it sees must be in shadow, and so not changed by being in shadow.

Expand full comment
TGGP's avatar

Your first link contains this passage:

"Finally, one of the most revealing tests involves Earth's shadow. No matter how asymmetric or irregular the distributions of plate defects may be, they have no plausible reason to avoid the Earth's shadow. In contrast, transients associated with solar reflections would. This shadow test provides a crucial empirical lever to distinguish between physical reflections and random defects - and remains an essential part of any validation framework moving forward."

It says nothing about what the nature of the things genuinely reflecting sunlight must be, nor does it say that the telescopes must always be looking into Earth's shadow (rather, the point of the analysis was that they were seen LESS often in Earth's shadow). If a different paper made that argument about Earth's shadow, then perhaps I overlooked it.

Expand full comment
Robin Hanson's avatar

To telescopes at night, everything in the Earth's atmosphere is in the shadow of the Earth. But only a small % of it is in Earth's shadow when projected out to GEO distances. So how would that % be suppressed in the data?

Expand full comment
Boltzmann Booty's avatar

These flashes occur tens of kilometers in the air

Expand full comment
Robin Hanson's avatar

Anything less than ~5000 km is too close.

Expand full comment
Boltzmann Booty's avatar

Again, what if the flashes are much less likely in the shadow, suppressing expected number compared to what you worked out?

Expand full comment
Boltzmann Booty's avatar

This diurnal lightning cycle seals the deal, atmospheric electrical discharges correlated with daylight are obviously the explanation https://www.researchgate.net/figure/ISS-LIS-diurnal-variability-of-global-lightning-flash-rate-including-land-ocean_fig9_343308598

Expand full comment
WallTapecapval's avatar

By that logic we should continue seeing them every year or see them even more with larger nuclear explosions. But instead we see that there is a dramatic drop in the occurrence of these glints post-1956, despite above-ground nuclear tests continuing, and even the worlds largest nuclear test (tsar bomba) happening in 1961.

Expand full comment
Boltzmann Booty's avatar

I don’t think there has been a drop in flashes, it’s more that there are way more glints now because there so much stuff up there. The flashes are interesting because they predate human satellites. But I at this point don’t think my suggestion works, because of the geometry

Expand full comment
WallTapecapval's avatar

despite the fact that we have much more things up in space now, it is still true that there was a dramatic drop in the occurrence of glints in 1956/1957 and the years after. Not an increase.

Expand full comment
Boltzmann Booty's avatar

What if presence of sunlight makes a given atmospheric phenomenon much more or less likely to manifest

Expand full comment
Boltzmann Booty's avatar

More specifically: Many fewer flashes in shadow than expected could be because the physics generating the flashes is stimulated by sunlight

Expand full comment
Boltzmann Booty's avatar

Even could be something as simple as: less vigorous convection during nighttime leads to less lightning + related phenomena

Expand full comment
Boltzmann Booty's avatar

Without a strong reason to reject this I think it’s clearly more likely than aliens

Expand full comment
Boltzmann Booty's avatar

“Sprites appear as luminous red-orange flashes. They often occur in clusters above the troposphere at an altitude range of 50–90 km (31–56 mi).”

Expand full comment
TGGP's avatar

This data all seems to come after the nuclear era. I'd like if we had pre-nuke data where we could exclude nukes as a cause to see if any of this phenomena still occurred.

Expand full comment
Robin Hanson's avatar

What could nukes possibly put up into high Earth orbit that would make flat reflecting surfaces?

Expand full comment
TGGP's avatar

I don't know, I'm just saying that pre-nuke data would exclude nukes as cause of any data seen at that time. If a nuclear government became dedicated enough to research this topic, I suppose they could conduct randomized nuclear tests specifically to examine this phenomena, but I doubt that would happen.

Expand full comment
WallTapecapval's avatar

These glints cease after 1956, despite nuclear tests (even above ground tests) continuing beyond then, and even the worlds largest nuclear detonation (Tsar bomba) occurring in 1961.

So that seems to rule out nukes as a cause.

Expand full comment
Lazar Ilic's avatar

Surely some physicist or astronomer will provide another explanation in this comment section? As an aside, the notion of a transient not only appears in some digital signal processing texts. But audiophiles and engineers often claim the Sony C800G handles certain events more sharply than a Warm Audio 8000. I knew a lad at CalTech who might quiz people on atoms, molecules, and wavefronts.

Expand full comment
Kevin's avatar

I’m not a physicist but I am working on radio telescope data analysis pipelines right now… the problem with interpreting transient blips in both radio and optical is that there are dozens of different sources of interference. The instrument itself often causes interference. Weather can cause interference. There probably isn’t just one cause for blips like this, there are probably many. There are dozens if not hundreds of papers which mention as an aside “oh also we saw thousands of unexplained little blips”. So you just can’t answer the question of “what is this” when you see some little thing.

Expand full comment
Xero's avatar
Aug 22Edited

Robin, thank you for the article. I’m not a mathematician. Just trying to reason this through as logically as I can. So please pardon my layperson’s interpretation.

I agree with the analysis. The shadow deficit is quite large and does rule out many trivial causes (as you have explained).

I am less sure about the conclusion you have drawn. Again, because the deficit is so large! I see two possibilities:

Either all of these transients were the same phenomenon (let’s call them 'aliens'), in which case we would expect a lot of independent corroboration from other surveys. And also, we would need to reconcile this with the lack of observations in subsequent decades when our instruments became increasingly more sensitive and broad. Meaning - thousands of alien probes appeared this one time, in a very specific orbit, and then were never seen again.

Or, most of these transients can ultimately be explained away by non-alien yet unknown causes (or errors), leaving maybe a handful unexplained. I recall the paper’s author suggesting in an interview that this is what she expects. But in that case, if we can plausibly explain most of these transients with still-unknown but non-alien reasons, then logically we should be able to explain all. There is nothing special about some of these transients compared to others because they all failed the shadow deficit test equally.

So my question is: would you agree with that framing? And if so, which position would you take?

Expand full comment
M.J. Jacobi's avatar

I've ran across people talking about this before, one suggestion was it's an artifact from the digitization of the original photographic plates:

https://academic.oup.com/rasti/article/3/1/73/7601398?login=false

Expand full comment
Robin Hanson's avatar

That can't explain the high correlation with the Earth shadow calculation.

Expand full comment
Christopher B's avatar

First time I've read about it but that explanation sounds a lot more plausible.

I get there's a lot more junk up there but at the same time it isn't like there aren't millions of people carrying around more processing power than we used to get to the moon in their back pockets now, and a heck of a lot more observing of stuff in reasonably close proximity to earth. You would think somebody would be talking about unexplained observations being recorded right now, not discovered from pictures that'll be a century old in the not so distant future. It seems a huge coincidence that aliens came by for a lookie-loo at just exactly the right time for us to record them on those plates and somehow have never been observed in any other fashion again.

Expand full comment
Phil Getts's avatar

The linking of the glints to the first nuclear tests makes the alien hypothesis much more credible, because it annihilates our priors for alien observation. Remember Arthur Clarke's short story in which an alien race set up an artifact on the moon to notify them when humans developed the ability to travel to the moon and make a nuclear explosion?

Expand full comment
Xpym's avatar
Aug 17Edited

Suppose that (presumably unfathomably advanced) aliens in fact are here observing us, but apparently avoiding interaction. Should we be doing anything differently?

Expand full comment
Thrawn's avatar

I'd think the way we treat one another would likely demonstrate to them something very important, so we might want to present better than we do today

Expand full comment
Xpym's avatar

We aren't being naughty out of sheer perversion though - by and large nobody is a villain of their own story, and given that the aliens haven't exterminated us in disgust already, this consideration probably won't move the needle much.

Expand full comment
Thrawn's avatar

Idk I feel even Robin is naughtier than he'd be if he knew for sure there was an intelligent alien species observing his views on gender.

Expand full comment
Xpym's avatar
Aug 17Edited

Eh, people used to believe that the literal omniscient God watches them all the time (and will horribly torture the naughty ones for eternity), and yet this didn't result in an earthly paradise...

Expand full comment
Thrawn's avatar

I don't think people really believe that it's just something people say they do. Regardless belief and knowledge can have different results

Expand full comment
Xpym's avatar

Nowadays probably not, but seems likely a few centuries ago.

>belief and knowledge can have different results

I don't think we'll get incontrovertible evidence for aliens any time soon, so "belief" will likely be the extent of it.

Expand full comment
AnthonyCV's avatar

There are a lot of things (we can all disagree on what exactly) we should be doing differently whether aliens are watching or not :-)

It's less clear whether those 'shoulds' differ much between the alien and non-alien cases.

Expand full comment
Jim McGinness's avatar

Interesting, but the evidence is far too weak to support a claim of alien artifacts.

Expand full comment
Jon Prescott's avatar

Jesus. That gave me chills.

Expand full comment
Robin Hanson's avatar

As well it should.

Expand full comment
Dave92F1's avatar

It aliens wanted us dead, we'd be dead. So either they don't exist or don't want us dead. Yet.

Expand full comment
Vasco Grilo's avatar

Thanks for the post, Robin!

"Thus at least ~2/3 of these ~100K 1950s transients are in fact brief glints off of flat shiny objects in roughly GEO! (See explainer in Added section below.)"

Beatriz Villarroel, the 1st author of the paper, says around 1/3 (https://youtu.be/hAAOSdvHtjA?t=688).

Expand full comment
Tim Tyler's avatar

High-altitude ice crystals causing bright flashes involving reflected sunlight is documented here:

https://news.agu.org/press-release/ice-particles-in-earths-atmosphere-create-bright-flashes-seen-from-space/

Apparently, the reported glints are from high-altitude clouds. This was widely reported in the science news in 2017. Here is a more recent article which offers a mechanistic explanation:

"Sunglints are made by floating, hexagonal platelet-shaped crystals that are about 1/10 of a millimeter large and form in freezing air. The crystals by nature of their form and physics tend to line up horizontally in the air, creating a sort of composite reflective surface that can bounce light like a mirror"

- https://www.earthdata.nasa.gov/news/feature-articles/glint-understanding

Expand full comment
Robin Hanson's avatar

High altitude clouds are way to close to explain the shadow correlation here. And those are way to small to explain the size of these glints.

Expand full comment
Tim Tyler's avatar

The ensembles of microscopic ice crystals line up: "tiny hexagonal platelets of ice, floating in air in nearly perfect horizontal alignment, are likely responsible for the glints observed by EPIC over land." The alignment effect can sometimes produce very large glints - as seen here: https://en.wikipedia.org/wiki/Sun_dog

As you say, atmospheric effects seem to not explain the "shadow" analysis in this paper. My inclination there is to defy the data - either the data or the analysis of it has issues. That seems more likely than there being a lot of flat surfaces near to a geostationary orbit. Or if there really are flat surfaces out there, they are a water / ice / crystal phenomenon - and not a field of engineered debris.

Expand full comment
Christopher F. Hansen's avatar

Even if UFOs exist and cause visual signals, it seems unlikely that they would do so at 1/5 the rate of all current human-launched satellites combined. Hence at least some portion of the pre-1957 signals are likely due to some other source, perhaps an unknown one not considered here. And if this source exists, it seems plausible that it's responsible for 100% of such signals.

Expand full comment
Phil Getts's avatar

"I’m told that by volume most ice in near Earth space is in ~1-2mm diameter flakes, and most of the count is in sub-micron specks, both of which seem way too small to cause these big glints."

I'm not familiar with the physics of ice freezing into planar crystals in space, but any process that forms a crystal should have a phase transition at the percolation threshold, beyond which the expected final crystal size is infinite. That means, I think, that the size of crystals will have a power-law distribution around the phase transition. "By volume most flakes are 1-2mm in diameter" implies that the system is usually below the percolation threshold, but changes in atmospheric conditions that move towards that threshold might produce a significant number of flakes with an 11-28 mm diameter.

That hypothesis strikes me as implausible--hard to believe the atmosphere is near a phase transition past which an ice shield might form around the Earth--but maybe as plausible as aliens. It's much more plausible if there are events like lightning strikes or nuclear explosions that move local regions of the atmosphere towards that phase transition.

Expand full comment
AJ Gyles's avatar

Amazing. I had no idea that telescopes had this kind of ability to see these small glints, let alone from high altitude and from the past.

Let's assume this is actually true. That raises the question... what would we actually do about it?

... do more above-ground nuclear tests to see if the UFOs come back?

... shut down satellites in a particular orbit to see if we can still see artificial glints there?

... start doing targeted radio broadcasts aimed at ETs?

... just ignore it and say it doesn't matter?

I really don't know what the play is here! It's crazy to think that UFOs could have been here, in a detectable way, and yet there's just nothing we can do with that information.

Expand full comment
JCambias's avatar

I would love it for this to be true, and I definitely think it deserves more investigation.

But. This does require our hypothetical aliens to be kind of careless and incompetent.

"We want to observe the Earthlings. Preliminary data suggests they are making use of electrical power and are experimenting with crude radio-frequency broadcasts. Detailed surface surveys indicate the presence of multiple large-aperture optical instruments for astronomy. We should therefore use stealthy stand-off observation platforms to reduce the risk of being detected."

"Nah, they'll never notice if we stick some cheap probes in synchronous orbit. It'll be fine."

Expand full comment
Robin Hanson's avatar

Maybe they want to be seen, but not clearly: https://www.overcomingbias.com/p/ufos-what-the-hellhtml

Expand full comment
Jack's avatar
Aug 18Edited

Interesting! I haven't a full reading of this, nor all of the references. However one thing jumps out at me. From page 13 of the preprint, regarding the set of 106,339 transient events selected for their study:

"An important note about the sample is that, contrary to the other transient candidates discussed throughout the paper, this sample has not been visually inspected. As such, it is expected to contain a substantial number of false positives, including clustered artifacts such as edge fingerprints or other plate defects that contaminate our sample."

It is remarkable that they haven't done a visual inspection of these events, nor provided summary statistics about them. Were they evenly distributed across the photographic plates, for example?

It's a clever idea to use Earth's shadow, and their conclusion rests entirely on that analysis. However the admitted unknown number of false positives, and with unknown distribution, are a real problem that the authors underplay.

Expand full comment
Robin Hanson's avatar

As no known errors could mimic the shadow process, the high correlation with the shadow calculation limits the rate of errors.

Expand full comment
Jack's avatar

One can imagine a scenario where (a) a lot of false positives cluster around the edges of the plates (due to fingerprints, scanning artifacts, etc.), and (b) the shadow images tend to undersample the edges for geometrical reasons. (E.g., a small circle drawn at a random location in a rectangle will tend to not intersect the edge.)

This would conspire to give the observed "deficit" of events in the shadow.

I'm not saying this precise mechanism is what's happening. Just that it's easy to construct alternative explanations like this when we know almost nothing about the overall dataset.

Expand full comment
Robin Hanson's avatar

"shadow images tend to undersample the edges for geometrical reasons". huh?

Expand full comment
Jack's avatar

Oops, my mental model was off. The umbral shadow at geostationary orbit subtends an angle of 19 degrees, which is larger than the 6 degree field of view of each plate. So for the most part a given plate is either in shadow or not in shadow.

Yes it's harder to make a confounding model in this case. Hmm.

Expand full comment
J.L.Mc12's avatar

I am pleasantly surprised that you have written an essay about this, because I had already heard about this phenomenon for the first time when I attended Jordan Shanks/ “FriendlyJordies” live show “Alien Hunter” a few weeks ago. It’s good to have some more in-depth writing about what he mentioned only briefly.

Expand full comment