Hacker News new | past | comments | ask | show | jobs | submit | gwern's comments login

I was reading through the references in https://blogs.umass.edu/comphon/2017/06/15/did-frank-rosenbl... linked in another comment, specifically https://gwern.net/doc/ai/1993-olazaran.pdf https://gwern.net/doc/ai/1996-olazaran.pdf by a Mikel Olazaran who spent a while interviewing what looks to be almost all the surviving connectionists & Minsky etc.

Olazaran argues that all the connectionists were perfectly aware of the _Perceptrons_ headline conclusion about single-layer perceptrons being hopelessly linear, which drafts had been circulating for like 4 years beforehand as well, and most regarded it as unimportant (pointing out that humans can't solve the parity of a grid of dots either without painfully counting them out one by one) and having an obvious solution (multiple layers) that they all, Rosenblatt especially, had put a lot of work into trying. The problem was, none of the multi-layer things worked, and people had run out of ideas. So most of the connectionist researchers got sucked away by things that were working at the time (eg the Stanford group was having huge success with adaptive antennas & telephone filters which accidentally come out of their NN work), and funding dried up (for both exogenous political reasons related to military R&D being cut, and just the lack of results compared to alternative research programs like the symbolics approaches which were enjoying their initial flush of success in theorem proving and checkers etc). So when, years later, _Perceptrons_ came out with all of its i's dotted and ts-crossed, it didn't "kill connectionism" because that had already died. What _Perceptrons_ really did was it served as a kind of excuse or Schelling point to make the death 'official' and cement the dominance of the symbolic approaches. Rosenblatt never gave up, but he had already been left high and dry with no more funding and no research community.

Olazaran directly asks several of them whether more funding or work would have helped, and it seems everyone agrees that it would've been useless. The computers just weren't there in the '60s. (One notes that it might have worked in the '70s if anyone had paid attention to the invention of backpropagation, pointing out that Rumelhart et al doing the PDP studies on backprop were using the equivalent of PCs for those studies in the late '80s, so if you were patient you could've done them on minicomputers/mainframes in the '70s. But not the '60s.)


This is made up out of whole cloth. I've never in my life seen a education analysis where C grades outperform the rest, give me a break.

Yes, after 18 years of WordPress development, the architects' excuse of blaming bad actors in plugin ecosystem has begun to wear thin, especially when contemporaries or predecessors like Firefox or Debian never had remotely the same level of problems with their users being hacked constantly by plugins/packages.

“No Way To Prevent This,” Says Only CMS Where This Regularly Happens

Firefox has had the converse problem of constantly breaking all their plugins by changing their APIs etc..

They were able to fix this by moving to the WebExtension model instead of letting extensions directly interface with XUL/XPCOM[0]. Of course, then everyone got angry that they couldn't do ridiculously invasive changes to the browser with an extension anymore.

[0] which were then massively refactored, twice.


Because it's not Waymo. It suggests that Waymo has a real competitor and Waymo-level tech has been replicated outside Waymo and their long lead over everyone else is evaporating because they have (to outside appearances) dithered so long. (One wonders if Waymo would even have 'been doing it for 6 months already' in SF if Cruise hadn't been ramping up in SF the past few years...)

It's kinda sad seeing them bought up again. I was reading https://www.vice.com/en/article/xwqjg3/the-complete-untold-h... yesterday (great but long read) and they really desperately wanted out of MS, and - victims of their own success with Halo 1 - made a lot of Halo games they didn't want to as the price of freedom. And now here they are again in a big conglomerate.

Curious. I wonder what Matplotlib does that's so bad? If it's merely being pathological, then compressing the PDFs with JBIG2 would make them potentially even smaller. (I usually do that with 'ocrmypdf', which doesn't have to OCR your PDF.)

Depends on your PDF and your goals, I think. JBIG2 is for raster data, but Matplotlib outputs vector images by default when producing PDFs. Rasterising may help with things like huge numbers of points that hide each other, but at the cost of sharpness and scalability.

InstructGPT was already submitted.

Sorry didn’t see it.

This apparently was a joke.

Yeah, in my family no one ever made a distinction. You'd look someone up in the yellow pages, they were all in the same stack, and only a prig would correct you, "you mean, look him up in the White Pages".

In the US, we said "the phone book" to be generic. The specific books were always specific.

As in "Did you look him up in the phone book?" which might be answered "I couldn't find him in the White Pages, but the phone's probably in his mom's name and I think she has a different last name, so that might be why. I think she has a business but I don't even know what industry so the Yellow Pages don't help here."


That past does seem to disappear in Google: https://news.ycombinator.com/item?id=23977375 https://news.ycombinator.com/item?id=19604135 Maybe people are evolving to erase dates as a countermeasure?

I sometimes wonder how much of this is just a ratchet of things like banning spam: on a long enough time horizon, the survival rate of everything goes to zero?


Applications are open for YC Summer 2022

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: