Feynman’s Maze-Running Story
gwern.net
gwern.net
Alternative hypothesis: Feynman knew who the individual was but, having not asked them if they objected to being used to illustrate his anecdote, chose to hide their identity with a pseudonym. In that case the reference wasn't to "Mr Q. Young", it was to "Mr Young Researcher".
I wonder if it's possible that there was a transcription error, or perhaps something introduced by the editor of the journal where the speech was first published [1]. If there was, then it's perhaps also plausible that the error could have been carried forward by copying and pasting the journal article into Surely you're joking.
All wild speculation on my part, of course. I wonder if any further evidence will ever turn up!
[1] https://calteches.library.caltech.edu/51/2/CargoCult.pdf
You're suggesting that _no one_ ever used a pseudonym for another without explicitly mentioning it in advance/shortly thereafter?
Besides, "Young" as a false-name doesn't have to be a direct clever "reference" to "a young person". It could just be a name he thought of at the time.
In any case, I think Gwern is far too conspiratorial in the reasoning for this article. There are some interesting connections found but the evidence, while reasonable, is not nearly as conclusive as he suggests.
In academic writing like this, where he is being quite precise about the names and explicitly stating 'a man named Young', where the speech is in fact about being precise & scientific, and the anecdote makes up a good chunk of the speech, which was written in advance, and published at least twice by him afterwards over a decade? Yes. Feynman would have highlighted or signaled a pseudonym, he would not have used a plausible surname (I will remind the reader again, quite a common one back then, and still is).
> the evidence, while reasonable, is not nearly as conclusive as he suggests.
You are suggesting that all this is a red herring and we will find another, completely independent, multi-year investigation into rats using floor cues in mazes done by American grad students in the late 1930s (when American academia was far smaller than it is now), published in a way Feynman could find out, at institutions with links to Feynman via Cornell & Michigan, which also chose to use sand for the same task and rationale, which is not cited by Shepard or Curtis or indeed by any researchers in any of the scores of papers I have jailbroken & read on the topic of sensory leakage in mazes, which has simultaneously eluded all discovery by at least a decade and a half of people taking cracks at it (the 2009 Quora, and then several blogs going back to 2004, not counting the '90s Usenet discussions), and which happens to by sheer coincidence, be the actual source for the anecdote?
That seems unlikely.
I suggest that Shepard had a net-negative effect on the field. Anyone who might eventually have done the same research would have heard about Shepard's work and been discouraged from re-doing it. Yet, it remained obscure and little-used. It may be better to not do the research at all than to do it, mark the territory, and not publish.
This meant that we had to acknowledge that the result already was claimed by someone else, but that we couldn't understand the proof attempt.
This of course leads to problems because now our result is somehow viewed as less valuable, and more difficult to publish. Which again means that research in that area will suffer from an unpublished claim.
I started doing research long ago enough that I got a bit of exposure to that world, in the sense that that publication process was still briefly in place, and I saw those white papers passed around and used. Some of them were very influential.
The problem in some ways is that as people migrated to the current system with an emphasis on published articles, that stuff that people were actually using kind of dropped out of consciousness, sort of along the lines the article is describing. I can think of unpublished white papers that were cited immensely; people made fun of them never being turned into published papers. Maybe it's hurt the the paper in the current day along the lines of the target piece, but what's sad or funny is it never occurred to people that there was a work that was being cited a lot that never got published, as in to question the value of the publication process itself.
I don't disagree with you but there's a bit of broader historical context that's missing from the piece I think.
The irony is that today we might be seeing a return to the white paper model, in the form of preprints and open access archival papers, but it's just massively distributed. Maybe there's some lessons to be learned from that lost previous era (Google search dropping links, etc.)? Maybe it's just inevitable that as time goes on some things will be lost regardless.
Were these nerds toiling to the benefit or tune of some charlatan, some _Trismegistus_ -- or were they too much the hermit, one who toils without the clutch or gearing for meshing with the gears of society? With the central limit theorem their sovereign then probably both I guess, instinct says more the latter.
The way you file the teeth of your gears is a real struggle. I wish there were more resources for it, AI has an opportunity to fill a gap here. I know the path of salvation is doing my normal do _just more aggressively and in public_.
I managed to carve out a real small niche making an indie game publicly over a year and a half, posted a few hundred videos. Some knocks and I lost the momentum for constant publishing. The work hasn't stopped, but the ephemeral threads of a narrative feel harder to grasp. Do I just start posting things out of nowhere, acting like I never stopped? Do I montage the unpublished backlog? Do I wait until I've achieved some scientific breakthrough?
I've got videos of flood fill lightning and neural networks and sheep shapeshifting and explosions and probably 100 other things I've forgotten. Maybe the problem is that I never knew who my audience was. Shouting into the canyon is harder than it looks.
Basically all of my textual content ends up here on HN :)
Knowing your audience is indeed step one. Truly.
Build at least one Ideal Customer Avatar [0]. Don't skip the step about visualizing them, even if it feels silly - it helps.
Maybe start a monthly newsletter, mostly so people are reminded you exist once in a while. Your unpublished backlog is probably a great resource for this.
And yes - just start posting again.
Good luck with the game, looks fun.
0 - https://gibbonet.co.uk/5-steps-to-build-your-ideal-customer-...
I think this is a good first step you've pointed me towards.
Edit: According to Caltech, they have "93 boxes" of Feynman's work that has never been backed up digitally[0]. Beyond the issue of access, what happens if there's ever a sprinkler burst in the wrong room of the archives? [0]https://collections.archives.caltech.edu/repositories/2/reso...
no, i never wore gloves. them was hard times!
Personally I'm very interested in the survivability of scientific information. I used to think that this was a worry of the Internet age, but papers locked up in working group reports and journals are increasingly hard to find beyond about a 25 year window. There was a thread on HN in 2020 where someone was looking for a NATO working group report from 1960 and you'd think NATO of all people would preserve records...but there's a black hole.
Edit: The article states B.F. skinner did create a box to remove these side-channel effects for rat experiments
How do we know they haven't?
And then there's this: https://esoterx.com/2014/12/03/murder-by-math-the-irrational...
There was institutional pushback on Godel's incompleteness from the axiomatic mathematicians that thought they could prove everything as they reconstructed math from axioms (if I understand that). From what I recall reading the axiomatic reconstruction was basically a huge industry of mathematical work, and suddenly some pissant young guy shows up and says "yeah, that not going to end the way you hope".
The mechanism of tenure alone will lead to political grandstanding. If your research is borderline on tenure but some other research completely undermines it, that is millions of lifetime dollars on the line.
See: MOND vs Dark Matter, string theory, etc.
I think it goes along with the Kuhnian (?) idea of paradigms becoming overly dominant and overly entrenched, and coming and going wholesale in waves. It's also consistent with Stigler's law of eponymy. As fields lurch from one idea or paradigm to another they tend to pretend that anything similar didn't exist beforehand. There's a need to represent ones's self and ideas as revolutionary and genius and implicitly, by comparison, to denigrate everything that came before as irrelevant.
This article describes a related but different phenomenon:
https://www.nature.com/nature-index/news-blog/review-article...
One of my curiosities is whether or not this has gotten worse in science and academics.
It often feels like in academics there's such a pressure for self-promotion and receipt of attention, to capture attention and be at the center of what is the focus of attention, that issues of veridicality or what has and should be the focus of attention gets swept away.
Well, I think there can an opposite problem that appears looking at these things. Psychology is such garbage, so likely to produce worthless experiments, that those that don't want to call a spade a spade instead try to take all the garbage of the field and average it over all science so that they can have something different to say.
This relates to the "replication crisis" [1]. You can say "The replication crisis is frequently discussed in relation to psychology and medicine..." but also said to extend throughout science. It may indeed extend throughout sciences but including a data group (experimental psychology) that you is garbage isn't necessarily a good way to see what's happening in the rest of science.
And, of course, a field "having amnesia" by itself doesn't necessarily imply any kind of problem. A given practitioner A) can only remember so much and B) benefits from a compact statement of a result, which may be quite different from how the result was originally stated. The math I learned in my MA came from text books that gave self-contained summaries of the various topics, rather than from the original papers the ideas came from and that made it much easier.