I hope to preprint my paper for your review on arxiv next week titled:
"A Novel Bridge from Randomness in Stochastic Data to, Like, OMG I'm SO Randomness in Valley Girl Entropy"
We will pay dearly for overloading that word. Good AGI will be capable of saying the most random things! Not, really, no. I mean, they'll still be pronounceable, I'm guessing?
You kind of have to read the article to understand the big bold headings.
He is postulating that the way we deal with physical memory (an example of L2 and L3 caches is provided) demonstrates that, as we proceed with trying to form AGI out of our classical computer architectures, there are some fundamental problems for example larger caches are slower. With human intelligence, this doesn't always seem to be a problem for some humans. If you understand the "attention" part of recent developments in this field, he's saying that transformers are the most efficient way we got to achieve that and it's starting to look like a problem from a "physicality" standpoint as the author puts it.
The "physically" here is that larger caches are not just computationally larger banks of data but are actually physically larger and, by Euclidean distance, further away. Yet paradoxically the elephant nor blue whale is not the smartest brain on the planet, the distance from the center of my head to broca's region seems to have no effect on my elocution. Etc. Studying Einstein's brain doesn't do much (I guess the insulators are somewhat important?) for understanding Einstein's intelligence ... but that is 100% critical when understanding L2 and L3 caches on a die.
> Who is to say biological computers won't be built?
No one is saying that. We're just pretty sure it's not happening in our lifetimes.
I think you misunderstand the author's intent. He's not saying "The things I can't imagine are not going to happen". He's trying to argue that, "Look, the way things are going, the diminishing returns we are already seeing, the way our hardware works, this isn't going to get us to AGI." Of course, if you had some new architecture or weird "wetware" that somehow solved these problems I'm sure this article would concede that that's not the point.
What if you took those scripts that you have used to automate your life, dumped them into something like cursorAI and asked the model to refine them, make them better, improve the output, add interactivity, generalize them, harden them, etc?
Sometimes when I have time to play around I just ask models what stinks in my code or how it could be better with respect to X. It's not always right or productive but it is fun, you should try it!
You do realize that before ansible there was a whole generation of scripters thinking "yay! some framework that is just going to be endless proprietary bugs I can't fix!"
I asked if you tried it, it sounds like you have I guess. I'm sorry you did not find another tool for your toolbox. I did.
You jest, but we praise Cern for inventing the web, not the millions of people who actually built websites and uploaded content. Someone who kicked things off via directing funding gets zero credit.
Which explains the politicians we end up with, it’s really kind of depressing.
So you're effectively arguing that quid pro quo isn't quid pro quo if you can't prove the counterfactual (that this contract would have happened without drilling rights)?
Doesn't that seem ... stupid? To put the onus on the people who have no access to any communications or details about what looks like corruption? It's my opinion that without forcing them to do everything in the open and prove to us that there is no quid pro quo, you're going to end up with Soviet USSR style government real fast.
I'm sure this is "business as usual" but ... maybe it shouldn't be?
It’s literally in the definition of “quid pro quo”: “a favor or advantage granted in return for something.”
By your line of logic, if Elizabeth Warren votes for something that benefits Harvard, that’s improper without any evidence that Harvard is giving her something in return.
It could be referring to the the underpinnings of how these things are used.
Use race as a dimension for something and that ends up as a value in a vector that packs a human into a discreet set of pigeonholes. Then take many of those and stack them and you've got a matrix ready for things like principal component analysis or CNN training.
You might say "oh come on, that hasn't been done since WWII by IBM" and you'd be wrong. It still happens today with things like calculating insurance premiums and approving bank loans. And your response might be "no way, nobody records someone's race" and while that might be technically correct, we frequently harvest things like income and interest in products that are highly correlated with spacefic races (some innocuous others much less innocuous). This can be harvested through cookies in websites like facebook or they can be self reported income on credit card applications.
You can disagree that it's the same as saying "matrix multiplication is racist" but that is just a boiled down way of saying "we are very good at hiding racism in our algorithms and then acting super surprised when someone points them out and our defense is that we just did some math."
This is a very poorly written article that myopically focuses on only the most extremely recent events in the culture war. Wouldn't a better article look at how wikipedia handles, say, the the civil rights act of 1964 and the rise of Evangelicism? Things that are, you know, many years ago and should be much more stable articles than these fly-by-night everyone is having a conniption over at this very moment?
Reading this article is kind of like I'm watching Jerry Springer except it's not as entertaining. As things stabilize the articles get better and as time stretches to infinity so too does the editing work eliminate bias on Wikipedia. Is it the greatest thing ever that you should unquestioningly read? Not even close. The author thinks it should be impossibly neutral. Nirvana fallacy.
Ah so newly created articles are "noisy" in terms of bias and over time they quiet down and approach a neutral point of view? if this is true, you should be able to show me a bunch of articles about polarizing current events that are heavily biased in the opposite direction to the ones featured in the article.
(I don't think this is possible, therefore I think you are wrong)
Nowhere did I claim that there were equally slanted liberal compared to conservative changes. Are we going to play the "reality has a well known liberal bias" game? Am I supposed to find articles on wikipedia that claim Biden lost the election or that no one should get a vaccine? I mean, this is a political statement: "Now that we have tons of vaccines, every death due to covid is a travesty." You know at least in the USA it's hard to find mainstream media that is conservative because those views have become so asinine. Is that my problem? Is that Wikipedia's problem?
What next? Are you going to complain that 98% of college professors are liberal? Okay, go change it. Enjoy trying to get people who enjoy liberal endeavors to promote conservative views. Am I shocked that the people who want to freely edit and debate articles on a website for no profit happen to slant liberal? Not really.
> (I don't think this is possible, therefore I think you are wrong)
What part exactly was I wrong about? I suggested you look at articles from 30-50 years ago and point out the liberal slant. Instead you invented a claim that I did not make so that I would be incorrect. Strawman.
If we cared enough about this, we'd all be reading conservapedia.com.
Honestly the section of the article on "The Antifa/BLM riots" was absolute garbage. He frames it that "National Democrats generally supported the rioters; portrayed them as “mostly peaceful” activists against fascism and racism, even contributing money to their defense; took seriously the notion that we should “defund the police” or backed similar police “reform” proposals; and stubbornly minimized the months of bloodshed, danger, and destruction the riots caused." But then he notes that at least one article included all this information and also agrees that the main article does a great job in neutrality. Then he has a problem with how monument removals was framed. sigh okay this guy is clearly cherry picking.
Trash article from someone trying to stay relevant.
Please provide the exact quote where he says the main article does a great job in neutrality. What I see is that he says there is good coverage of relevant facts but the way in which the article writes about them is obviously biased.
> A neutral treatment would, of course, give broad factual coverage of such things as where the rioting took place, how many people were arrested, and numbers of injuries and deaths attributable to the rioting. The main Wikipedia article actually seems to do a good job there, as far as I can tell.
Sorry he said "good job" and I said "great job". I suppose there is a difference there and I apologize for that.
Why don't you quote his article and then the wikipedia "interpretation" that is "very biased"? Let's take concrete examples. I'll start, here's an excerpt from the section in question in Larry Sanger's article:
> The rest of the article—which, I confess, I did not read entirely, as it is very long
He cares so much about the bias that he can't be bothered to find it. The only thing I saw (and I mentioned in my original post) is that he took issue with the description of statue removal phrasing. So he skips through three long articles, doesn't read them all but finds what he's looking for. This is largely how people operate today, they start with what they want to believe then they go looking for it.
First, your statement that he "cares so much about the bias he can't be bothered to find it" is objectively false. He found two examples in the summary, which is what 80% of people will read, and will anchor interpretation for all readers. Bias in the summary is much more significant than bias in the main article, even if it returned to even-handed interpretation later.
Also to your point: "This is largely how people operate today, they start with what they want to believe then they go looking for it."
If he went looking for it, he certainly found it very easily. I would agree with you if he had found the bias in niche topics and in later sections of articles much less likely to be read. But when the articles with the most traffic and most contributors demonstrate bias, its clear there is a problem.
Again I requested concrete examples. Please post what he said and what's actually word for word in the summary you are claiming is biased. It's really easy to talk abstractly and say whatever you want. I think you'll find that once you try enumerate this, it's hard to defend Sanger's assertion that the articles are clearly biased towards liberal positions.
> Why don't you quote his article and then the wikipedia "interpretation" that is "very biased"? Let's take concrete examples.
That's a request for concrete examples. Regardless, why don't you provide concrete examples of the liberal bias in those articles to back up your claims?
Since I know you to be a respectable and upstanding member of society, I assume all of your requests come with the word "please." Or at the very least "I request."
Your use of the word "Let's" was either a suggestion or an order. Based on your tone it appears to have been an order XD
Nowhere in the article does it mention the cost to the healthcare system per innoculation. It's also an article written to make it sound like Merck just wants to save lives but you need to also consider the billions off dollars this would net them. Also, buried in the article was this gem:
"Though most experts have little doubt that the vaccine prevents HPV infections anywhere in the body and therefore the resulting cancers, technically, this has not been proven."
This has been extensively studied, and virtually every analysis comes to the same conclusion: it's cost-effective.
Here's an old analysis: https://www.ncbi.nlm.nih.gov/pubmed/12533280, which puts its effectiveness on-par with MMR vaccines and cost-effectiveness significantly better than the median $/QALY for existing interventions (~$42k, I think).
Well...Suppose the vaccine did nothing at all when administered to males but--for some reason--we did it anyway. This would double the price while leaving the effectiveness unchanged, but the $/quality-adjusted year of life (QAYL) would still be a in range that we typically consider worthwhile.
In reality, the vaccine has beneficial effects for the men themselves, as the article points out. It also contributes to "herd immunity", which benefits vaccinated men and women alike. Increased demand for the vaccine could also drive down the unit cost (most drug-related costs are fixed, rather than marginal). These three factors all suggest it's even more cost-effective than my naive analysis above.
Also, definitely talk about stuff like this. It's important and interesting; I'm actually surprised there's less interest in the pharmoeconomic modelling itself, as it's a very HN-friendly topic. That said, it is a little frustrating when people assume everyone involved in something biological is an innumerate dummy. A lot of this stuff has been analyzed, often by fairly clever people.
Odd that you would suggest the article is written to show Merck in a good light, and then choose that exact quote to make your point.
To give the whole quote in context:
"Merck, its maker, wouldn’t be likely to mention the potential benefit in advertisements. Though most experts have little doubt that the vaccine prevents HPV infections anywhere in the body and therefore the resulting cancers, technically, this has not been proven."
It's not exactly calling Merck out, but the author is clearly not giving them a pat on the back over it either.
In the UK, the NHS's NICE organisation casts a fairly critical eye over cost effectiveness of medication. All girls &and boys are now offered the HPV jab at 12-13 years.
By contrast, we don't routinely vaccinate against chickenpox.
I wouldn't put much concern on that not proven thing. Most vaccines aren't technically proven to work in the way we think about proving things in other areas of science. For starters, vaccines are not required to go through double blind studies. We give you the vaccine and if antibodies show up we say it works.
That isn't the problem. The problem is it takes too long to run this hypothetical study from "Vaccinate 12 year old boys" to "These 50 year olds didn't die of cancer at the expected rates". Nobody wants to wait 40 years to run a study that says this thing we suspect is a good idea is, in fact, a good idea. 40 years is a long time. If you follow a patient for six months, they're easy to find. Maybe a handful of them move city once, but they remember they were in a medical trial, "Oh yeah, sorry, I will come in for the follow-up appointment". If you try 40 years it's hopeless. They move ten times, they leave the country, they've forgotten all about your medical trial. So you start with 100 patients and 40 years later you can only trace eight of them, and now your trial is statistically invalid.
The existing efficacy demo for HPV vaccination against cervical cancer relies on the fact that Pap smears are a thing to get answers in just a few years not decades. The smear test gets us a bunch of cells that a pathologist can look at under a microscope and inspect to see if they're normal or maybe "pre-cancerous" meaning they aren't normal but they are not yet cancer. And we know (from having tried it) that if we just ignore it some of that not-yet-cancer turns into cancer and kills people. So if we see that in a smear today you'll get called back and they'll treat it, even though you haven't got cancer.
This means they could give girls an HPV vaccine and then not only show that those girls didn't get the strains of HPV vaccinated against as often as unvaccinated peers, but further show that they don't get as much not-yet-cancer. And _by implication_ that means it protects against cancer.
There is no smear test for these less common cancers in men. If there are pre-cancerous cells we don't see them. We only find out about the cancer, years and often decades later.
I think the "anywhere on the body... technically, this has not been proven." is also an infelicitous bit of wording. The links to anogenital cancers have been studied pretty thoroughly. There's less data on the head/neck/throat ones and virtually none at all on the vaccines' possible effect on warts elsewhere on the body.
It's presumably all the same mechanism whereby HPV infection stresses cells and eventually causes cancer, but it's much further from "proven".
"A Novel Bridge from Randomness in Stochastic Data to, Like, OMG I'm SO Randomness in Valley Girl Entropy"
We will pay dearly for overloading that word. Good AGI will be capable of saying the most random things! Not, really, no. I mean, they'll still be pronounceable, I'm guessing?