probably_wrong

> The thinking guy with the one approach will likely lose a photo competition to one of the accidentally better shots of the person who tried a lot more.

If that were the case then photo competitions would have been upended by now by tourists who accidentally took masterpieces with their phones.

I can take as many thousands of pictures as I want but, without taking the time to critically review them to see what works and what doesn't, I will rarely produce a prize-worthy picture. And then there's the risk that I'll discard it anyway because I don't have the eye to separate the prize-winning picture from the Instagram shot all my contacts are publishing.

First, to get it out of the way: I'm seeing comments here that clearly haven't read the article. I encourage you all to do that.

I feel the article gets it right when suggesting it's the amount of (not always accurate!) data they can throw at you. In an honest discussion there's an assumption that the other person won't straight up lie to me so if someone shows me ten examples for why my argument is wrong I may be inclined to believe them. But if half of those examples are made up, well, that's a different story.

I still think of the commenter here who said "LLMs are a DDOS on free resources" and I feel the comparison works here. If police officers can overwhelm innocent people into confessing, then so can an LLM that "can't be bargained with, can't be reasoned with, doesn't feel pity, or remorse, or fear! And it absolutely will not stop, ever, until you are"... convinced.

x and y are common shorthand for inputs and outputs respectively. Saying "y=LLM(x)" denotes that the input tweets (x) are fed to a black-box function (LLM) that returns a classification (y) for that input. Given that we're talking about binary classifiation it's expected for y to be a value between 0 and 1, which leads to defining LLM(x) as "a function that receives a tweet x as input, uses an LLM, and returns a value in the [0, 1] range representing its classification as ironic or not".

The point about beta (which I think the equation doesn't actually reflect) is just to indicate that this approach risks nothing because, worst case scenario, the weights you assign to the model can simply return the original LLM prediction. Keep in mind that 0*inf=0 and that LLM(x) only returns 0 or 1, so it's not accurate to say that the sigmoid would always return 1.

The function I() is the indicator function [1] which, in this case, returns 1 if the argument is true and 0 otherwise. It's only there to convert booleans to integers because summing booleans is not defined.

[1] https://en.wikipedia.org/wiki/Indicator_function

> How else are they to monetize text outputs?

Sam Altman famously said in 2024 "I kind of think of ads as a last resort for us for a business model." So perhaps we could ask him first what his other plans were and why they didn't work.

The fact that ads are one of two types of content for which people regularly install blockers (with porn coming second) speaks volumes about how useful they really are.

> Because what else can you do?

You can resist.

Upvote quirky projects, call out fake hype, make your disapproval visible, call AI slop what it is.

The term AI has become a liability in some circles, so let that be proof that you don't have to just accept whatever AI companies tell us.

I was wondering this myself so I wrote a script that shows that there are duplicated puzzles if you count a rotated Sudoku as being equivalent. Here's one example:

Puzzle 2 is

  1234
  3412
  2341
  4123

If you rotate it counterclockwise you have

  4213
  3142
  2431
  1324

And if you normalize it, replacing the first row with "1234" (brilliant idea from the post), you get

  1234
  4312
  2143
  3421

Which is listed as puzzle 7. A quick check gives me that puzzles 1, 2, 3, 5, 11 and 12 would be unique under rotation, but I wrote that check in five minutes so there must be bugs somewhere. Also, I performed no check for mirror symmetry whatsoever (update: I did a quick one, same result).

I don't want to miss my chance to say that the post is brilliant and that it convinced me to leave what I was doing to check for rotations. I was nerd sniped in the best way and I take my hat off for the OP.

There's no need for a new crime when we already have reckless conduct, namely, "conduct that creates a substantial and unjustifiable risk of harm to others and involves a conscious disregard of, or indifference to, that risk".

https://www.law.cornell.edu/wex/reckless

The explanation I got from my doctor was that one shouldn't fix posture problems by raising the inner part of the insoles a random amount because all that achieves is making you run "weird", putting pressure in the wrong part of the articulation and damaging the ligaments that keep the knee in place. The actual solution involved muscle development exercises, ripping out the insoles of those shoes to achieve a flat base and replacing them with custom-made ones, knee pads with metallic inserts, and a two-year training plan.

Semi-related: I did the gait analysis thing when I started seriously training for a half marathon. The guy showed me the video and reached the conclusion that, since I run with my knees flexed inwards (true) I needed a certain shoe that would "stabilize" my feet and push my knees outwards.

It took me several years and several pieces of orthopedic equipment to recover from the damage those shoes did to me. Do not trust salesmen with medical advice.

No. You will have your slop and you'll like it, too.

I couldn't finish the book because my Kindle died halfway, but based on the opening chapters I agree with you.

I admire her for the stance she's taking today and the personal sacrifice it's costing her to this day, but I won't deny that her own account sounds like someone who thought "this will be big" and decided to help without thinking "should this be big?" first.

Perhaps that's the way to go: if we build our social media in agent-only forums no one will mine it for training. The only downside will be the inevitable "Are you a robot?" captchas.

I often read here comments along the lines of "there will be plenty of jobs for AI-unfuckers" [1], but I fear the truth may be closer to "sometimes things are broken, are then fixed with a broken solution, and will remain broken till the end of time".

In my last round of job search this year no one asked me questions about good software practices but they did ask me how proficient I was with AI agents. And let's not forget that we are part of an industry whose motto is "move fast and break things". Sure, NASA has a zero-bugs approach to software dev, but they don't have room for all of us.

[1] https://news.ycombinator.com/item?id=44646820

I'd like to see a citation for "most of the west".

I'm going to disagree with your law of "conservation of filth" on the grounds that it's pushing things too far. Whatever amount of feces comes into your system via walking downtown on a regular day and then bringing the shoes inside pales in comparison to the amount of fecal matter that enters via your toothbrush if you keep it in the bathroom. At some point a small amount of dirt is just fine, as evidenced by my immune system not even visibly reacting whenever I pick up and eat something that fell on the floor (as long as I follow the five-seconds rule). And people do go out to restaurants or, even worse, picnics, where all that separates their food from a much filthier surface is a piece of cloth.

I'm not going to say that every type of floor is to be walked on, with rugs and tatami mats being perfect counter-examples. But if a five-star restaurant doesn't ask me to take my shoes off (not evening Japan, AFAIK) then keeping them on is, from a sanitary point of view, probably okay.

Coming from a shoes-on culture I can give a couple points.

First, in the rare cases you describe (when was the last time you stepped on feces?) one would indeed leave their shoes outside. But in the day-to-day your shoes have some dust on them and that's it - you won't even notice it's there when compared to the dust that already accumulates daily. If you live in a climate where this assumption doesn't hold (snowy areas come to mind) then the equation changes.

Second, the floor is usually stone and therefore designed for being walked on with shoes. I would feel awkward walking on some polished wood, but if the difference between inside and outside is "this one is less coarse but just as hard" then I wouldn't see a good reason for taking my shoes off. I would probably avoid the rug, but that's me.

As for comfort, keeping your shoes on also means one less thing to do when going in and out. Do not underestimate the path with less friction.

There's an article somewhere (read it many years ago) with the story about the first days of Tinder. The way they got their first users was to send (IIRC) the marketing strategist (a woman) to talk to attractive young women in bars and parties, show them how easy it was to use, and convince them to install the app themselves.

I would say that this points to the real issues: that OkCupid was something one did on their desktop at home while Tinder sold itself as something one does for fun at the bar, and that most people are laser-focused on looks in detriment of everything else. Last time I used Bumble (which was supposed to be "deeper" than Tinder) I'd yell in frustration, scrolling profile after profile without even the tiniest bit of information, eventually reaching the conclusion that people just don't want to read or write profiles.

A new OkCupid would have to position itself as the "eat your vegetables" alternative in a world full of candy. There is a market for that, sure, but it will remain a niche at best.

I clicked on a couple ads once because I was desperate - I needed a cheap flight with a tight deadline and my usual search engines didn't help. I thought "if these ads deliver what they promise, I'll never book anywhere else".

Believe it or not, they did not deliver.

For starters, it breaks two rule of the thread: "Agencies, recruiters, job boards, and so on, are off topic here" and "please only email these addresses to discuss work opportunities".

Not that you don't get spammed anyway, but still.

Mildly related side note: my Kindle DX died this year (it won't charge nor reset, even with a new battery) so I've been looking for a replacement. I'd like a second-hand one but, based on my eBay experience, barely anyone is getting rid of their e-readers.

Considering how many second-hand computers, tablets, and cell phones are there, I find this surprising.

The problem of "what actually happened" is that we don't have enough information to properly understand what happened, what's new, and what's not.

We do have language to talk about emergent behavior, with "evolutionary algorithm" being the first one I'd expect in a serious discussion. And we do have mechanisms for algorithms to coordinate with each other using language, as seen in my above-mentioned Facebook experiment from 2017. But instead of writing "our evolutionary behavior encodes state in the first-available memory position which is then reused by subsequent clones" which would properly focus on what's new and what isn't, we are talking about conspiracies and "the Philip of Macedon of this second AI civilization". Even the METR report (which is miles ahead of this article) argues that they had to use unreliable AI in their conclusions because they had six days to analyse 1300 chains of thought and 70000 messages.

I would love to talk about the science behind this experiment. A PR piece is not helping with that.

Do you remember that time in 2017 when Facebook reportedly shut down AIs after they started "talking to each other in their own language" [1]? Instead of reporting the story as "we set the parameters for our optimization problem wrong and we had to stop it because it overfitted", the press went with a version of "AI is going to kill us all".

This article feels exactly like that: by intentionally using human terms like "civilization" or "brotherhood" the article is deviating from what actually happened to present a story about how AI is all but alive. I'll go ahead and predict that this story will be remembered the same way as that one other scientist who argued, in 2023, that Google's AI was alive [2].

[1] https://www.independent.co.uk/life-style/facebook-artificial...

[2] https://futurism.com/blake-lemoine-google-interview

If a fireman is trying to stop a house fire and suddenly they catch fire themselves, no one would argue that they need to focus on the house fire first.

Antidepressants are the same: they won't stop the bigger fire by themselves, but by preventing you from catching fire right now they give you the tools to deal with the larger problem more effectively that if you tried to do it while flailing around and screaming.

I'm not convinced by the author's point that ACs do not impose a burden on the electrical grid. They write that ACs have gotten more efficient, which may be true, but that's the same Windows developers have been saying over years and yet boot times and hardware requirements have gone way up. One million efficient units still require more energy than no units. And while I definitely believe that European hospitals and retirement homes sadly need ACs now (thanks global warming), I'm not sure everyone and their dog needs one.

I also have reservations about why an American libertarian think tank writes about what Europeans should do, but that's probably on me.

Having seen the increase in suicide rates for girls and young women [1], and having read the arguments for why no other explanation fits that data, I am convinced social media is indeed to blame.

But as someone else put it once regarding global warming, what's the worst case if we reign in social media anyway? That we make the Internet a less addictive place for nothing?

[1] https://www.economist.com/graphic-detail/2023/05/03/suicide-.... Archive version: https://archive.is/wY1OH

Whenever I need something from Google Books I inevitably reach the message that this is a limited preview and the part I need is not included.

I therefore feel the same way about Google Books that how I felt when I learned that What.cd went down: that I don't gain or lose anything anyway because I never had access to begin with, and that by not making it 100% publicly accessible you're asking for the data to one day disappear forever.

I know Meta it's not Google, but it's worth remembering that these promises haven't had a great measure of success in the past:

> Facebook has been fined €110m (£94m) by the EU for providing misleading information about its 2014 takeover of WhatsApp. (...) When Facebook took over the WhatsApp messaging service in 2014, it told the commission it would not be able to match user accounts on both platforms, but went on to do exactly that.

https://www.theguardian.com/business/2017/may/18/facebook-fi...

It has a 1-to-1 correlation, though: the faster you can generate code (which is the one quality under discussion), the more lines of code you can generate in a given unit of time, and vice versa. So I would argue we are talking about lines-of-code, just indirectly.

I don't see how one can hold "speed of code generation" as a good metric while, simultaneously, holding "number of lines generated in a unit of time" as a bad metric. And more worrisome, I'm not sure how one would reward one without rewarding the other creating a perverse incentive. But I'm open to be proven wrong.

Lines-of-code has long been accepted as a terrible measure for code quality, and I believe it has been strongly correlated with poorer code quality.