Priors and Surprise
post by MichaelVassar · 2010-03-03T08:27:53.554Z · LW · GW · Legacy · 32 commentsContents
32 comments
I don’t want to be too dogmatic about this claim, but Godzilla is unrealistic. I don’t want to be too non-dogmatic about this claim either. OK then, just how dogmatic should I be? I have all sorts of reasons for thinking that skyscraper sized lizards or dinosaurs don’t actually exist. Honestly, the most important of these is probably that none of the people who I imagine would know if they did exist seem to believe in them. I never hear any mention of them in the news, in history books, etc, and I don’t see their effects in the national death statistics. No industries seem to exist to deal with their rampages, and no oil or shipping companies lose stock value from lizard attacks. Casually, at least, Godzilla attacks don’t seem like the sort of basic fact about the world that people could just overlook. How confident should I be that Godzilla type creatures don't exist?
I can also fairly easily recognize good biological reasons not to expect there to be giant rampaging lizards. The square/cube law, in its many manifestations, is the most basic of these, but by itself is not completely decisive. I can imagine physical workarounds that would allow sequoia giganticus sized reptiles, but not without novel bio-machinery that would take a long time to evolve and would surely be found in many other organisms. I can even vaguely imagine ways in which biology might prove resistant to conventional military weaponry and ecological niches and lifestyles that might support both such biology and such size, though much of my knowledge of Earth’s ecosystems would have to be re-written. For all that, if I lived in a world where essentially all authorities did refer to the activities of godzilla giganticus I would probably accept that they were probably correct regarding its existence. What should a hypothetical person who lived in a world where the existence of Godzilla type creatures was common knowledge and was regarded as an ordinary non-numinous fact about the world believe?
Godzilla would be considerably more perplexing than thunderstones, and would have to be considerably better documented to be credible. Even with the strongest documentation I would have substantial unresolved questions, inferring that Godzilla’s native ecosystem must be quite different from any known (possibly inferring that the details are classified), and even wondering whether Godzilla was a biological creature at all as opposed to, for instance, a giant robot left behind by an advanced and forgotten civilization, a line of inquiry that would greatly increase my credence in secret history of all kinds. For the most part though, I would probably go about life as normal. Even Natural Selection, the most damaged part of my world-view, would endure as a great intellectual triumph explaining the origins of almost all of Earth’s life forms. Only peripheral facts, such as distant history and the nature of some exotic ecosystems would be deeply called into question, and such facts are not tightly integrated with the broader edifice of science. In a conversation with a hypothetical Michael Vassar who believed in Godzilla, the issue would typically not come up. Science in general would not be called into question in my mind, but should it be?
32 comments
Comments sorted by top scores.
comment by Eliezer Yudkowsky (Eliezer_Yudkowsky) · 2010-03-04T19:14:13.677Z · LW(p) · GW(p)
What is the point of this post?
Replies from: spriteless, MichaelVassar, Clippy↑ comment by spriteless · 2010-03-05T20:48:06.797Z · LW(p) · GW(p)
'How would I explain it if I wake up and my left arm is replaced with a tentacle?' and ' How would I explain it if Godzilla attacks?' are similar questions. Except Godzilla is more on your mind, because no one makes a movie about how my arm was replaced by a tentacle, but it's easy to come up with drama about a giant fire snorting lizard.
Replies from: StartAtTheEnd↑ comment by StartAtTheEnd · 2023-11-25T05:55:13.301Z · LW(p) · GW(p)
because no one makes a movie about how my arm was replaced by a tentacle
https://en.wikipedia.org/wiki/Parasyte:_Part_1
You're probably not on here any more, I just though your comment aged badly in a comical manner
↑ comment by MichaelVassar · 2010-03-05T21:10:51.120Z · LW(p) · GW(p)
Calibration and the proper roles of revision of hypotheses vs. replacement.
Replies from: JGWeissman↑ comment by JGWeissman · 2010-03-05T23:38:30.459Z · LW(p) · GW(p)
Although I see where you talk about calibration and choices between revision and replacement of hypothesis, I did not learn anything new about these subjects from the post.
Replies from: MichaelVassar↑ comment by MichaelVassar · 2010-03-06T00:48:38.417Z · LW(p) · GW(p)
Fair criticism. It's hard to know what the quality threshold for posting should be. Ideally bad posts would net negative karma for feedback purposes but people are unfortunately reluctant to downvote.
comment by Morendil · 2010-03-03T08:58:40.335Z · LW(p) · GW(p)
wondering whether Godzilla was a biological creature at all as opposed to, for instance, a giant robot left behind by an advanced and forgotten civilization
That's one of my main take-aways from PT:TLOS so far. (This post has reminded me of one of the exercises in the book.) The idea in Chap. 5 that given extremely unlikely priors for a hypothesis, strong evidence appearing to favor that hypothesis will not so much confirm it as call for the the resurrection of another, highly unlikely, but still more plausible than the original.
"You are being deceived" is advanced as a useful hypothesis to consider resurrecting in general. Perhaps conspiracy theorists are not as stupid as they sometimes appear; if you think the straightforward hypothesis unlikely enough, mass deception can be a more appealing hypothesis, with readily available mechanisms of explanation. (For instance a movie like The Truman Show couldn't work if we didn't assign a small degree of plausibility even to mass deception, given sufficient financial incentive.)
Replies from: GreenRoot, Hook↑ comment by GreenRoot · 2010-03-04T17:45:33.236Z · LW(p) · GW(p)
For others' reference, PT:TLOS = Probability Theory: The Logic of Science, by E. T. Jaynes.
↑ comment by Hook · 2010-03-03T15:01:04.610Z · LW(p) · GW(p)
Jaynes even uses the example of meteorites (aka thunderstones) to show that this line of reasoning, while valid by the laws of probability, can lead educated people to believe things that are not true.
The natural philosophers who knew something about gravity had a much higher prior probability for the unreliability of farmers' reports of natural phenomena than they did for rocks falling from the sky, so every report served to reinforce the hypothesis that farmers don't know what they are talking about. It took reports from people who were considered reliable in order for the meteorite hypothesis to be accepted.
How reliable a reporter would you need to call science in general into question?
comment by Karl_Smith · 2010-03-03T18:52:36.832Z · LW(p) · GW(p)
Perhaps I am missing something but it seems to me that a world in which Godzilla was common knowledge would have a completely different history of biology. For one thing it's hard to imagine that explaing Godzillia would not be a major goal of philosophers and scientists since the earliest days.
I imagine one of the basic questions would be whether Godzillia was a beast or a god and answering this would be a high priority. What does Godzillia want? Where did he come from? Has he always existed? Are there more? Do they mate?
These seem like really big deal questions when confronted by a sea monster which occasionally destroys towns.
Replies from: Strange7comment by MrHen · 2010-03-03T14:50:31.816Z · LW(p) · GW(p)
EDIT: For historical purposes, this comment reported two typos that have since been fixed. I was intending to delete this comment when they were fixed but a valuable discussion occurred below.
Replies from: Hook↑ comment by Hook · 2010-03-03T16:23:44.955Z · LW(p) · GW(p)
As a suggestion, maybe typos that have no substantial impact on readability should be communicated to the author through a direct message rather than a public comment.
Replies from: Clippy, JGWeissman, Vladimir_Nesov, MrHen, Morendil, Richard_Kennaway↑ comment by JGWeissman · 2010-03-03T18:23:54.250Z · LW(p) · GW(p)
I disagree. Public comments have the advantage that other readers who noticed the typo will be aware that it has already been reported, so they don't have to bother. And that the public record of the typo report and the author's response saying it is fixed encourages people to report such errors, leading to more readable articles for future readers.
Replies from: MonBonify↑ comment by MonBonify · 2010-03-03T23:37:38.017Z · LW(p) · GW(p)
I find it distracting when people report typos in the public comments--more distracting than the typos themselves in the actual post. There should be a better interface that: 1) allows users to easily report typos without writing a comment or a direct message 2) drives awareness that the community should help edit posts 3) alerts users that a typo has been reported in a unintrusive way.
Perhaps they can make each line of a post a live link that you can click to view a pop-up box in which you can write an edit. A little red dot in the margin by a line in which a user has reported an error would be enough to let people know it's been taken care of. A little note in the margin that explains the system will help spread the word.
In lieu of a new system--we have to figure out if it's more important to influence the readability of articles or more important to insure the flow of comments is not disrupted. I would argue that the comment flow is way more important, considering how the caliber of the users in this community protect the readability of the articles from any serious jeopardy. The flow of comments, on the other hand, is what makes this forum uniquely nurturing to rational thought. This detailed matrix of connecting ideas is as reflective of the cognitive process as it is supportive of it.
Replies from: FAWS, Larks, Larks, Larks↑ comment by FAWS · 2010-03-04T00:19:15.720Z · LW(p) · GW(p)
That sounds like it would require considerable programming effort without all that much gain. Do you volunteer to do it yourself?
Replies from: MonBonify↑ comment by MonBonify · 2010-03-04T15:48:42.437Z · LW(p) · GW(p)
Yes, I volunteer myself. I would need feedback on the best solution--as the one I previously outlined was just one way it could be done. Right now, below each post are the following options: Vote up, Vote down, Comments (#), Save, and Report. They could easily add "Mark a typo" or "Report a typo" that could pop out a new window in which you can alert the author of a typo that needs fixing.
In terms of effort vs. gain--you pose an interesting question. I would argue that it is worth the effort. This is a website about rational thought, so it seems fitting that it should have a smooth user interface that they're constantly trying to optimize. I also think any effort into finding a better way or creating a better system is always well spent because it creates an environment that nurtures innovative thinking and progress. So even though programming a new function would only have an incremental benefit to users, I think little things like this have a positive impact on the way people think. Web programming is such a flexible medium, allowing improvements like this--and just like people on this site put extra time and effort into the pursuit of truth and rationality, I think we should also put extra effort into optimizing our systems of interaction, making them as efficient as possible. Efficiency is the sister of rationality.
Replies from: Hook↑ comment by Hook · 2010-03-04T16:36:37.551Z · LW(p) · GW(p)
You should consider other solutions, since the first one you think of is unlikely to be the best/cheapest to implement. The "Edit" functionality already exists. Users above a certain karma level could be allowed to edit posts, as in the case of StackOverflow. The major cost is that there would need to be a way to revert changes to prevent vandalism. Morendil pointed out that DM are a bit harder to send than comments. If desired, that could be fixed cheaply. There are surely other solutions.
Replies from: MonBonify↑ comment by MonBonify · 2010-03-04T17:48:23.141Z · LW(p) · GW(p)
The new functionality wouldn't allow users to edit the post, but rather alert the author that there is a typo that might need fixing--does that help clarify my previous comment? I agree that allowing users to edit posts without the approval of the author could do more harm than good.
Imagine that there is another link below each post that opens the text of the post in a new window in which you are able to highlight typos (this could be programmed in a variety of ways--I would want to do more research on it to determine the best one). Once submitted, highlighted typos will automatically be Direct Messaged to the author in a automated format (see example below) and a little red dot will appear in the margin next to a line for which a typo alert was submitted--helping make sure authors are not flooded with alerts about the same typo. If a dot in the margin is too hard to program, here is an alternative: When a user clicks on the link to create a typo-alert, the highlighted text from previous typo-alerts submitted by other users is displayed in the pop-up window.
i.e.
Dear [Name]
User [Name] has found the following typos in your post. [copy of sentence with highlighted text from typo-pop-out box inserted here]. You can correct any typos in your post by using the Edit function.
Thanks, The Typo-Alert Generator
This may not be the best system, but through collaboration I think we can figure something out. Until there is some kind of efficient system in place that leverages the fact that web programming can be easily changed to suit a site's specific needs, I think we should just ignore typos and allow the highest level users to edit any glaring mistakes for clarity as they see fit.
Replies from: Hook↑ comment by Hook · 2010-03-04T19:11:17.443Z · LW(p) · GW(p)
That sounds like a decent solution. I have no idea how hard the little red dot would be to program, but I think it would be distracting for the people who don't care about the typos. The highlighted text from previous typo-alerts makes sure that only the people who care get the information.
↑ comment by Larks · 2010-03-04T15:01:19.583Z · LW(p) · GW(p)
Report a typo, and then reply to your own report. As a footnote to the typo-report, you would ask people to downvote the report so it became hidden, and upvote the daughter-post as a karma-balance.
e.g.
Larks: You spelt ### wrong. NB: could everyone please downvote this post so it becomes hidden, and upvote the post below as a karma-balance. If you would like to express your dislike of this mechanism, simply omit to upvote the balance.
Larks: Karma Balance post
↑ comment by Larks · 2010-03-04T15:00:06.915Z · LW(p) · GW(p)
Report a typo, and then reply to your own report. As a footnote to the typo-report, you would ask people to downvote the report so it became hidden, and upvote the daughter-post as a karma-balance.
e.g.
Larks
You spelt ### wrong. NB: could everyone please downvote this post so it becomes hidden, and upvote the post below as a karma-balance.
Larks
Karma Balance
↑ comment by Larks · 2010-03-04T15:02:46.588Z · LW(p) · GW(p)
Report a typo, and then reply to your own report. As a footnote to the typo-report, you would ask people to downvote the report so it became hidden, and upvote the daughter-post as a karma-balance.
e.g.
Replies from: prase, MorendilLarks:
You spelt ### wrong.
NB: could everyone please downvote this post so it becomes hidden, and upvote the post below as a karma-balance. If you would like to express disaproval of this post, simply omit to upvote the balance.
Larks:
Karma Balance
↑ comment by Vladimir_Nesov · 2010-03-04T18:22:13.846Z · LW(p) · GW(p)
The best solution would be for the posts to be in a wiki, with two versions (a working version, and a public version approved by the author or authorized users) to keep vandalism from the front page, but with editing allowed to all users. This way, both typos and formatting issues could be resolved without bothering the author or anyone else (remember the encoding bugs in OB-imported posts, that nobody bothers to fix even after they are reported, not for a long time anyway!). Not a viable short-term solution, obviously, but technology will probably get there in a few years.
↑ comment by Richard_Kennaway · 2010-03-04T16:01:38.843Z · LW(p) · GW(p)
I would prefer that such typos simply be ignored.
comment by LauraABJ · 2010-03-06T23:05:43.622Z · LW(p) · GW(p)
We discussed a similar idea in reference to Godzilla, namely what kind of evidence we would need to believe that 'magical' elements existed in the world. The point you made then was that even something as outside our scientific understanding as Godzilla would be insufficient evidence to change our basic scientific world view, and that such evidence might not exist even in theory. I think this post could be easily improved by an introduction explaining this point, which you currently leave as an open question at the the end.