Decide Now; We Won’t Know Much More Later Re UFOs

By now humanity has seen hundreds of thousands of UFO reports, accumulated over more than eight decades. What have we learned about aliens from them? I suggest that we think about this as a four stage learning and analysis process. All four stages can be done at all times, and we do better over time by improving all of them.

First, we combine the evidence of specific reports, and our cosmological priors, to estimate a chance that each report so far report was caused by an encounter with a strange advanced civilization (wherever from), i.e., real “aliens”, as opposed to human mistakes, illusions, hoaxes, or foreign mischief. This estimation is done jointly, and produces a joint distribution over these chances.

Second, we define interesting categories of reports (e.g. saucer shaped, had big acceleration, reported by pilot) and then aggregate those report chance estimates across reports into the chances per category that any report in that category was from a real alien. Including a total chance for the total category of all reports. (And maybe also an expected count of alien encounters per category.) Now we have a total chance of any alien encounters, plus a joint distribution over the interesting categories of alien encounter events, conditional on any of them being alien encounters. E.g., most alien encounters tend to be sauce-shaped and to do a big acceleration.

Third, we combine those chances per interesting category with our best theories of cosmology, biology, and sociality, to infer probability distributions over aliens’ features, e.g., abilities, plans, origins, motivations, and other activities, conditional on any of these reports coming from a real alien. E.g., they have traveled to here to convince us to not expand into the universe, by being visible and impressive but saying little else, and stand ready to exterminate us if necessary.

Fourth, we combine these alien feature distributions with our preferences over various possible resulting outcomes, to consider many possible actions that we could take regarding aliens. Like to maybe broadcast more to them, spend more to defend against them, or spend more to search for where they are hiding. Arguably the main reason to do the first three stages above is to get better inputs for this stage of decision analysis. After all, in standard decision theory the only point of having more info is to help you make better decisions.

The above learning and analysis process has been going on for many decades, and should continue for many more. As we get more reports, and learn more about cosmology, biology, and sociality, and as our preferences change, we slowly adjust our estimates of the total chance that any encounters were of aliens, of our conditional distributions over categories of alien encounters, and of our estimates of the expected value of various related actions, including of key meta actions that might get us more info to better estimate the value of other actions.

In the last few decades, the UFO/UAP community has focused strongly on steps one and two, getting better data on particular encounters and categorizing them, and largely tabooed steps three and four, inferring alien features and estimating the value of related actions, due to a strong desire to be taken seriously, and a (reasonable) fear that steps three and four would make them look silly. They strongly want a higher consensus estimate of the total chance that any encounter has been with aliens, and to get credit for having helped cause this. They have been ridiculed and assigned low status based on claims that this chance is very low, and so a high chance would vindicate their efforts and greatly raise their status.

They also modestly want to better know the distribution over alien encounter categories. They are curious, after all. But for now they do not want to be associated at all with anyone trying to do steps three and four. And so little relatively effort has gone into those steps.

However, this strong focus on this status-pivotal total chance estimate seems to me quite excessive relative to the goal of better informing our key related decisions. Yes it has some relevance, but it seems to me that the conditional distribution over the interesting categories of alien encounter events is far more relevant for these decisions, as also are our changing preferences and our changing theories of cosmology, biology, and sociality.

Furthermore, it seems to me that these other more relevant considerations just aren’t changing that much over the decades. And they probably wouldn’t change that much even over a period when our best estimate of the total chance that some encounters were with aliens changed from 10% to 90%!

In such a period, we would gain enough added confidence in a few particular encounters, that they really happened the way that they seemed to, and that they really had features that seem quite hard to explain as human mistakes, illusions, or hoaxes. But they most likely wouldn’t much change the overall distribution of plausible encounters, and thus not change our inferred distributions over alien features, nor our estimates of the value of various actions in response. For example, we are quite unlikely to record a conversation with an alien, or to read explanatory documents written by them, or to photograph their home world. We’d mostly just be adding to the same sort of indirect evidence that we have now.

Thus our analysis of the expected value of various related decisions would be in the rough ballpark of being as good at the start of that period, when our total chance estimate was 10%, as at the end, when our total chance estimate was 90%. And we would have nearly as good a reason to do that sort of decisions analysis at the start of such a period as at the end.

Here’s the bottom line: After at least eight decades of seeing hundreds of thousands of UFO reports, we are nearly as well placed now to put serious effort into analysis steps three and four above as we will decades from now. Even if after those decades the world has come to admit that yes we have most likely been visited by aliens, and all those UFO folks deserve our deep apology, and new high status respect. The world might then be more open to listening to decision advice re aliens, but we won’t actually be much better placed to give it.

If the possibility that some UFO/UAP encounters are with aliens is actually important, because there are actually big decisions to be made that depend on this possibility, then we shouldn’t wait to do steps three and four above until many decades from now when UFO/UAP people have finally achieved the (likely deserved) respect that they crave. As we already know now nearly as much as we would then, we should do and publish those key decision analyses now as best we can. And then improve them slowly over time. So that the world can respond well to, well, aliens.

And if we just don’t know enough now to useful inform any related decisions, we should admit that we also won’t know enough for that decades from now either, and thus admit that all this UAP stuff is only being studied to satisfying some curiosities, and not because it will actually helps us make important decisions.

Added 21Aug: See here re the frequent false claim that we don’t have more better UFO pictures over time.

Added 22Aug: So far, none of the 37 comments are on the specific post topic, namely, if it is time to do steps 3,4, as we won’t know much more about them later. It really is true for some topics X that mentioning X in any way attracts the entire discussion to the usual debates about X, ignoring whatever was said specifically about X.

Added 24Oct: I’m feeling this all the more at the moment, seeing the glint results published. I’ve offered my best judgements and explanation of the evidence here.