ppod

Good points, but I feel like the real test would be if this was an unverified anti-Trump story. Did twitter or facebook put the brakes on any of the Steele dossier stuff?

The principle of putting the brakes on unverified stuff is good, but, of course, you have to decide what counts as "unverified", and you have to do that objectively, and quickly. What would the NYT say if there was an unverified anti-Trump story a week before the election, but twitter slowed the spread of it until it could be confirmed?

I love Spacy, but I was recently trying to embed documents with it and it seems that it uses word-vector averaging for that. Is it possible to do sentence or paragraph embedding like sentenceBert?

The cradle rocks above an abyss, and common sense tells us that our existence is but a brief crack of light between two eternities of darkness. Although the two are identical twins, man, as a rule, views the prenatal abyss with more calm than the one he is headed for (at some forty-five hundred heartbeats an hour). I know, however, of a young chronophobiac who experienced something like panic when looking for the first time at homemade movies that had been taken a few weeks before his birth. He saw a world that was practically unchanged — the same house, the same people — and then realized that he did not exist there at all and that nobody mourned his absence. He caught a glimpse of his mother waving from an upstairs window, and that unfamiliar gesture disturbed him, as if it were some mysterious farewell. But what particularly frightened him was the sight of a brand-new baby carriage standing there on the porch; even that was empty, as if, in the reverse order of events, his very bones had disintegrated.

--- Vladimir Nabokov

The rarely seen positivist-luddite niche.

I think parts of this message are very important, but the presentation makes it seem less interesting than it is. At first it looks to be just another "Social Dilemma" style "tech evil Zuckerberg bad" clone. Then it makes a connection to global warming, which, though it might be accurate, I think is an unfortunate link because that issue is more controversial than it should be.

The difficulty is that unless you make these kind of platforms illegal, people who engage with them in ways that enhance their reputations will have an advantage over those who choose not engage with them. And most people will always choose to use them, not only because they're convenient, but because people love an opportunity to enhance their own image in public.

For these reasons, I think it won't be possible to convince most people to exercise their right to privacy. What we should do instead is try to make society as tolerant as possible, so that there is no penalty for how you present yourself online. Urging people to exercise more privacy in fact has the opposite effect, because we hear fewer diverse viewpoints, and those who exercise their privacy come under greater suspicion.

Possibly a dumb question but why has RAM plateaued in high-end laptops? Very expensive laptops these days seem to have either 8 or 16GB of memory, and it seems to me as though this has been the case for about 8 years. Is there some reason that you don't see a benefit beyond that? Is it to do with using SSDs?

Wow very cool. It was the text I was interested in and I can see that it's tokenized in the XML, that's great.

"White men are so fragile," she fired off, "

Not everyone would see that as anti-racism

Is the underlying text collection available for bulk download or api query?

Mindchanging market's tough. Now the choir, the choir dollar's a good dollar.

Also there is an underlying class thing going on with "you are the product". Choosing to pay for a service is not just an equal alternative to a free service for most people. Most people have to work hard to afford paid services, many of them at low wages. Sneering at people who choose to use a free service rather than sell their time for $15 an hour is not a great look.

I'm not sure that those links completely show that you're wrong. The example you're replying to appears deeper than it is because each nested element is quite broad, so the absolute depth is not as much as it might seem (I think).

> We don’t do this by putting the two words in a sequence, like an artificial intelligence or a bonobo would do. Instead we build a new hierarchical unit. This unit puts together the verb drink with the noun wine to create the phrase drink wine, with wine being the grammatical object of drink.

AI can do compositionality.

>When users are scapegoated, Silicon Valley is left off the hook

Is this really the dominant narrative? Are there lots of thinkpieces going around about how fair and kind Silicon Valley is, if only it weren't for all those mean people using their services? Maybe it is so obvious that it doesn't need to be said, but when comparing the early internet to the current internet, we cannot let the users off the hook!

The further back you go, the harder it was to get on the internet, both as a user and a publisher. That meant the early internet was full of people who worked at universities, or were so motivated to discuss weird hobbies and interests with others that they struggled through the expense and technical difficulties, maybe even self-hosted, learned weird new languages from scratch. These were interesting, educated, intelligent, thoughtful and passionate people.

It's like air travel. If you were flying in the 1970s you probably had an interesting or important job or story. Getting on a plane meant meeting a bunch of interesting people. Now it doesn't. That is not a bad thing and it's not Southwest's fault. And when people complain about it, even though they might target Southwest (or Facebook), there is really an underlying snobbery about it. I can't believe I have to sit next to all these commoners on my internet!

I just noticed recently that in American English "bullshit" often means "unfair" or "trivial" where in British English it usually means "untrue".

What are the best third party reddit apps?

Interesting choice on one of those charts to truncate the "post-crash Bitcoin" price at end 2018. The price recovered to $10k by mid-2019 and has been relatively stable there since.

Absolutely. I'm totally on board with what they're trying to do, sounds like a great idea, but

>The mass surveillance and factory farming of human beings on a global scale is the business model of people farmers like Facebook and Google. It is the primary driver of the socioeconomic system we call surveillance capitalism.

This puts me off right away. I probably partially agree, but this kind of simplistic resentment is just a turn-off for the service.

Overall maybe, but I'd say the jury's out on the flight model vs X-Plane, and it (currently) doesn't model airliner systems quite to the level of the payware X-plane airbuses or the Zibo boeing. It really crushes X-Plane on weather currently.

The link to the "complete list of the experiments" is actually much more than that. It is a description of their methodology, and it's very revealing.

>These experiments are not, by any means, either a representative or a systematic sample of anything. We designed them explicitly to be difficult for current natural language processing technology. Moreover, we pre-tested them on the "AI Dungeon" game which is powered by some version of GPT-3, and we excluded those for which "AI Dungeon" gave reasonable answers. (We did not keep any record of those.) The pre-testing on AI Dungeon is the reason that many of them are in the second person; AI Dungeon prefers that. Also, as noted above, the experiments included some near duplicates. Therefore, though we note that, of the 157 examples below, 71 are successes, 70 are failures and 16 are flawed, these numbers are essentially meaningless.

https://cs.nyu.edu/faculty/davise/papers/GPT3CompleteTests.h...

There are already several comments here that put the word "understanding" in quotation marks or italics. It is beginning to be used in the same way that "consciousness" used to be used, as a kind of ill-defined catch-all for something that separates humans from machines.

Yes, there are clearly failures in reasoning, binding, and coherence in many of the examples here. There are many other cases where it does ok with simple reasoning tasks, maintains cohesion over many paragraphs, and successfully creates formal or generic text such as poetry, code, stylistic imitation.

I don't think that everyone who does research in the field would agree with your comment, or the article. More and more often I see people saying "real researchers in the field" know that GPT-3 has no understanding or reasoning ability, but I know people researching in the field who disagree with that.

I meant the OP IEEE article that these comments are on.

All I said was that as far as I know it isn't fully known. The final report isn't out yet. I'm just an aviation enthusiast, but it's notable that in many discussions anything that could possibly be perceived as even possibly implying any pilot fault is kind of taboo. I'm not saying the pilots did anything wrong, I'm just saying that leaving the stab cut-out switch out of the story seems strange to me.

It seems to me that this danger receives a much larger degree of attention than other dangers which are more likely to happen, such as road traffic death.

This article makes a really good case overall, but it must be deliberate that it chooses not to mention the stabiliser trim cut-out switch. In the part about the Cessna he says:

>There are instructions on how to detect when the system malfunctions and how to disable the system, immediately. Disabling the system means pulling the autopilot circuit breaker; instructions on how to do that are strewn throughout the documentation, repeatedly. Every pilot who flies my plane becomes intimately aware that it is not the same as any other 172. This is a big difference between what pilots who want to fly my plane are told and what pilots stepping into a 737 Max are (or were) told.

Which really implies that no such procedure existed for the Max, and that's just not true. There may have been problems with the procedure, as far as I know it's not fully known yet why that procedure didn't prevent the accidents, possibly inadequate pilot training.

https://www.flightglobal.com/safety/trim-cut-out-puzzle-emer...

Great article. I just wonder if this bit is really true:

>Most of those market and technical forces are on the side of economics, not safety.

People seem to think about airline safety a lot, like, an irrational amount given the risk. I would guess that over the decades consumer pressure was a huge driver of safety improvements, and the increasing automation seems to have been a big safety benefit on balance.

This reminds me of a paper from a while ago that used the sound of a person typing to get their password. "acoustic keyboard emanations". It seems to be a more convincing paper than the OP

https://www.cs.cornell.edu/~shmat/courses/cs6431/zhuang.pdf

I have no problem with the idea of down-weighting the entire national cohort to correct for this. The problem is that they actually weighted the results on a per-school basis.

Just to be clear about what actually happened, the results were weighted on a per-school basis, based on results from the previous four years. The disparity in weighting is not solely down to effects at the top of the distribution.

>Student populations have a fixed number of grades awarded every year, etc.

Yes, for the national cohort. This is different, the predicted results were weighted on a per-school basis. For example, a particular child taking, say, a history exam, in a school which had previously had no pupils that achieved A* in history over the past four years had probably no chance of getting an A*, even if that was their teachers' prediction. Purely on the basis of which school they went to. A weak cohort (compared to previous years) at a good school would be upweighted.