Have you considered using quickcheck/random/property-based testing with LLM code generation to automate function implementation?
3,688 karma · joined May 13, 2011
Anthropic said that mass surveillance was per se prohibited even if the government self-certified that it was lawful.
Anthropic said that mass surveillance was per se prohibited even if the government self-certified that it was lawful.
ps aux | grep nginx | grep root | grep -v grep
ps -u root -C nginx
psc 'process.name == "nginx" && process.user == "root"'These are what are claimed to be the onerous regulations slowing down innovation.
Copying and pasting tends to be a very tedious operation in the shell, which usually requires moving your hands away from the keyboard to the mouse (there are terminals which allow you to quick-select and insert lines but they are still more tedious than simply pressing enter to have the command on the line editor). Maybe try using llm-cmd-comp for a while.
> I do not see how editing the command is a good tradeoff here in terms of complexity+UI.
I don't find it a tradeoff, I think it's strictly superior in every way including complexity. llm-cmd-comp is probably the way I most often interface with llms (maybe second to basic search-engine-replacement) and I almost always either 1. don't have the file glob or the file names themselves ready (they may not exist yet!) at the time when I want to start writing the command or they are easier to enter using a fuzzy selector like fzf 2. don't want the llm to do weird things with globs when I pass them directly and having the shell expand them is usually difficult because the prompt is not a command (so the completion system won't do the right thing).
But even in your own demo it is faster to use llm-cmd-comp and you also get the benefit that the command goes into the history and you can optionally edit it if you want or further revise the prompt! It does require pressing enter twice instead of "y" but I don't find that a huge inconvenience especially since I almost always edit the command anyway.
Again, try installing llm-cmd-comp and try out your demo case.
- it puts the command in the shell editor line so you can edit it (for example to specify filenames using the line editor after the fact and make use of the shell tools like glob expansion etc.)
- it goes into the history.
- It can use a binding so you can start writing something without remembering to prefix it with a command and invoke the cmd completion at any place in the line editor.
- It also allows you to refine the command interactively.
- corfu+marginalia+vertico+embark+orderless is the standard completion stack now
- magit (maybe also see also the "casual" the transient front end for other modes)
- avy
- (fset 'yes-or-no-p 'y-or-n-p) ; 'Y' or 'N' instead of 'Yes' or 'No'
- (setq confirm-kill-emacs 'y-or-n-p)
- evil (optionally if you like vim keybinds, you still need to know basic emacs))
- if using evil: evil-collection, evil-args, evil-goggles, evil-traces, evil-escape, evil-nerd-commenter, evil-lion, evil-surround, etc are not "standard" but still useful
- configure melpa source
- which-key
- helpful
- undo-fu
- gptel
- projectile
- eglot
- saveplace
- desktop
- uniquify
- dired/wdired
- flycheck/flymake
- treesitter
- general (for making your own keybinds)
- some kind of multicursor mode (I use iedit but it's simple)
- yasnippet
- org (usefulness depends on the person)
Also, very important, learn the help system (C-h <key>), especially C-h f, C-h k, C-h w, C-h c. And the info system
1. quick-select the output in the terminal (select files/paths, urls, etc and either copy or paste them at the cursor). This is very useful as you often don't have the foresight to pipe the command to pipe the output of the command to some selection mechanism and even placing text on the line-editor is not that easy by default.
2. view images (sixel or kitty protocol). This is pretty useful visual analog to cat that doesn't require opening another program and works over ssh. Also for video.
There are some other nice utilities for doing things like downloading files directly in the terminal (it2dl for iterm and kitten transfer for kitty).
kitty doesn't work out of the box on macos if I remember; you have to set configuration for option/command etc.
For reference Enron was 40x earnings for current year forward estimates early in the year of the crash.
Why do you say that?
M-x sql-sqlite
Then C-c C-c on paragraph you want to send/execute
M-x sql-send- for the other commands
The mode is called sql-interactive-mode
This end-run around foreign government monetary control has been touted by Stripe executives as one of the main selling points for USD stablecoins but I don't see how foreign governments don't clamp down on this is in the same ways the clamp down on other uses of USD in the country; most monetary transfers have some physical presence or touchpoints the government can control.
More importantly the US itself is eventually going to come to the conclusion that it does not want people holding US dollars for similar reasons: it also loses control over monetary policy, with excessive inflows un-intuitively leading either to unemployment or excessive debt (c.f. Michael Pettis)
That said, it's possible stablecoin networks succeed for other reasons, particularly having a widely-accepted "API" that is developed at the pace of modern technology companies instead of laggard banks.
<new-landings.txt parallel --delimiter '\n\n' --colsep '\n' --plus curl '{1/large/medium}' -o '{#} + 101'.jpg
<new-landings.txt parallel --pipe --recend '\n\n' --colsep '\n' --plus curl '{1/large/medium}' -o '{#} + 101'.jpg
Adding multiple commands should be in a script to avoid quoting issues.
These are all movies with (at the time) >90% "approval" rating but average score about 7/10 with most reviews around the 6/10 threshold and tapering down at 7/10,8/10 (as opposed to being multi-modal/split-opinion, e.g. many at 6/10 and many also at 10/10).
If LLMs were already a breakthrough in proving theorems, even for obscure minor theorems, there would be a massive increase in published papers due to publish or perish academic incentives.
High chance given that this is the same guy that came up with SVG unicorn (sparks of AGI) which raises the same question even more obviously.
I literally provided several examples of unreasonable/misleading scores (there isn't an API so this is the best you can do). You on the other hand haven't provided any examples to show when RT is more useful than MC. The context of the original criticism is that the RT score is misleading and less useful than an actual rating.
> This is just gatekeeping masquerading as an argument. They have more to say on the mechanics of film, not on what movies you'll think are the best. It's especially ridiculous when you realize that RT gets reviews from these 'top critics' and you can filter for them.
It's not gatekeeping to recognize that there are actual experts in matter of taste who are going to be way more informative than the average joe. The implication of your argument is that every jack-asses opinion is just as valuable. No one actually believes this. And I don't know how many times I need to say this but I don't want to read a dozen critic reviews, I want a summary statistic (so that you can get a rough ranking of movies).
> The RT percentage has a critics and top critics score so no it's not, not anymore than the metacritic score is 'the score the average person will give it'. That's not how statistics work.
The problem is that the summary statistic (even if the average score was still visible) on RT includes so many people that it is closer to average opinion than critic opinion. That's apart from the problem with hot-or-not binary classification.
> There's nothing to pretend. If i think it's better then it's better for me. What critics say doesn't matter. It's really sad that you apparently characterize movies you like with lower critic scores as 'junk food'. Have a mind of your own.
I have explicitly said that I don't dogmatically follow critic opinions (no one does). The point is the opinions have value as a starting point to filter/select films that is better than popular opinion or a binary thresholding type system.
Once again you can make your same argument against any rating system (IMDB, RT,e tc) and absurdly claim that if you use those as a starting point to select films that you don't "have a mind of your own". In fact you can do this for ratings of any product or service and make the absolutely trivial point that a person's personal preferences or situation can deviate from consensus expert rankings. It's silly to then conclude that one ranking system can't be better than another or that popular opinion is equally valuable compared to expert opinion.
The original point of contention was that the "percent that approve" of the film that RT uses is surprising and not as useful as a regular rating system. (By the way the average score is now hidden on RT).
> So your argument against a broad-sample aggregator (RT) is to use a slightly-less-broad-sample aggregator (MC)?
My argument is to use useful aggregation of experts instead of a much less useful one.
> You complain about "every-person-with-a-substack" but you're still relying on an aggregation of dozens of critics you've never heard of.
I don't need to have "heard of them" to know that the NYT film critic, the reviewer at Siskel and Ebert and the film critic at Vanity Fair are probably more worth listening to than the "MacGuffin or Meaning Substack".
> You're just drawing an arbitrary line in the sand and calling it "quality."
No, the opinions of the film critics for the top publications in the world are not arbitrary.
> You're confusing your personal, subjective viewing threshold with an objective measure of quality. A score is what it is. 75/100 is the top quartile. That is, by definition, a high score. Whether you have enough time in your life to watch every movie in the top quartile is completely irrelevant to the validity of the score itself.
Beside the fact that the rating isn't the percentile ranking of film the entire point of the discussion is which site better helps you choose films. Again the definition of "high score" is completely arbitrary and irrelevant.
> Now this is more besides the point but i really do think that you're using a tool (Metacritic's average score) for a job it wasn't designed for: being the sole arbiter of what's worth your time. A film's "quality" is not a single, objective number. It depends on genre, intent, and audience.
I never said that. It's a helpful filtering mechanism. I watch low rated films if they are a genre I particularly like (just like I eat junk food without claiming that it is haute-cuisine) and I don't watch movies if they are not in a style I enjoy. Apropos of your example I don't like horror so I don't watch it, irrespective of the score.
> Not to mention your method is overly biased towards one specific type of film: the prestige drama. If that's the only kind of film you like to watch then cool i guess, but if not then what you're currently doing is nonsensical.
Most films are dramas as far as I know. In any case you can filter on categories so it's irrelevant.
> The RT score tries to answers the question..."What are the odds i'll like this movie?".
Well it's closer to what are the odds the average person will like it, which isn't what I want: I want 1. to be able to pick a better movie rather than a worse one and 2. be able to threshold on higher quality than the average person.
> This is a very useful information to have especially because i'm not a zombie picking movies to watch because of a single average score from an echo chamber of critics (which is bizarrely what you seem to be doing).
No one is doing this, they are using Metacritic as a starting point to filter and rank movies which, once again, RT doesn't do a good job at because of it's binary classifier system and inclusion of everyone under the sun.
> If the synopsis of Bob Trevino is more interesting to me, I would absolutely pick it over Past Lives. They are complementary scores.
That's fine and as I said something I and everyone else does, just like I eat junk food (and maybe sometimes actually prefer to some 3-star Michelin restaurant). The problem is pretending that those two films are roughly the same quality, or that because someone sometimes prefers a lower critic ranked movie that ratings don't matter: you can make the same argument about preferring a "rotten" RT movie.
It's much quicker and easier to just get an aggregated Metascore, which takes a second (and allows you to go in blind). I don't have any desire to read 12 movie review articles for every movie ever released.
> The entire purpose of an aggregator is to gather a wide sample to smooth out individual biases.
The point is to get a useful number not to achieve some platonic ideal in statistics. Again there aren't 100 movies critics worth listening to and I am not looking for popular opinion. If you want popular opinion use IMDB ratings.
> This is a nonsensical argument. By this logic, if we have a phenomenal year for film where 100 movies get a score over 75, the score itself becomes less valid? A score's meaning is relative to the scale, not the number of films that achieve it.
Yes is some fantasy world where that happens you would be right. In the real world that doesn't happen. Even if it did happen many people still have time constraints and want to watch only the best X films a year and Metacritic is just better at doing that than Rotten Tomatoes is. As yet another example "Bob Trevino likes it" is 94 RT vs 70 MC compared with "Past Lives" 95 RT vs MC 94: Which is more informative when selecting a movie? I can list more examples but I can't find any examples that demonstrate the reverse (i.e. that shows that you would be better off listening to RT over MC).
> And Literally hundreds of movies are released every year. 8 a month is a tiny fraction of that. Your personal viewing capacity doesn't change the fact that 75/100 is objectively a high score.
"High score" is an arbitrary definition. For the purposes of the discussion, which is whether Metacritic is a better way to determine which movies to watch, 74 doesn't cross the threshold of worth watching (absent some other factor) unless you watch more than 8 movies a month (and only want to watch movies released this year).
> No, you've asserted that. We've established they measure two different things. RT measures consensus (% of critics who liked it). Metacritic measures average intensity (a weighted average score). Both are useful. One tells you how many critics would recommend it, the other tells you how much they recommend it, on average. Claiming one is "not very useful" is just stating your personal preference as well as demonstrably false, as rotten tomatoes is very widely used.
Again, it is not useful in the sense of choosing movies to watch if you are even mildly selective. I gave another example above showing why. It's true that many people don't care about that, they just want something that the average person finds entertaining for 1.5 hours, and Rotten Tomatoes is fine for that. If you have a quality threshold higher than that or would rather watch a better movie than a worse one then it isn't.