jbstack

There is actually scientific evidence that direct engagement with material (e.g. making notes, re-writing in your own words, completing exercises, explaining it to others, etc.) is very beneficial to memory formation.

So, although creating a card is similar to reviewing it once (in that they will both help you remember it for a while), the former is worth more than the latter as a "unit" of memorisation. This means that you'll likely have longer review intervals, and therefore spend less time on reviews, if you wrote the card yourself, because the memory starts out stronger.

That has to be balanced of course with the amount of time you spend writing the card vs the gains you make in saved review time from having done so.

The problem with that approach is that adult second-language learners generally don't put in the same amount of time or access the same breadth of material as a baby learning it's mother tongue. Every word has a frequency with which it appears. A child will come across almost all words a sufficient number of times to eventually memorise them, because they are totally immersed in the language every hour of every day. As an adult learning a second language, unless I'm living in the target country in a fully immersive way (which for me isn't the case), then by consuming 30-60 minutes of media a day (which is my upper limit of what I can realistically achieve) I'm going to get strong exposure to high-frequency words and grammar patterns, and weak exposure to low-frequency ones. Many of those at the bottom end will be so weak that the occasional exposure I get simply isn't enough for me to attain fluency with them. Anki solves this problem: all words you learn via flashcards have a (roughly) equal chance of being remembered, independent of their frequency.

Anki also allows you to take long breaks from learning. If I go a year without learning any new material, provided I keep up my reviews (which significantly diminish in duration the longer you go without new cards), I'll pretty much be able to pick things up again where I left off. That doesn't work so well with other methods because you will forget a lot.

Flashcards are a lot more efficient in terms of number of minutes spent per word. For example, Skritter tells me that I spend an average of 1.78 minute learning how to write a word with Chinese characters. Aside from the fact that I wouldn't practice writing at all if I just consumed books/films, I'd also spend a lot more time that way as I'd constantly be stopping to look characters up in a dictionary and/or googling the grammar points every time I forget.

I take your point that you find Anki boring but that's highly subjective. I actually find it very satisfying and rewarding, almost as if I'm downloading information into my brain Matrix-style (just slower). There's a sense that whatever knowledge I put in Anki is mine to keep forever. In ~10 years of language learning, Anki is the one thing that I've most consistently kept up with. Your claim that it's a "weird cultural instinct [to make everything] as dry and uninteresting as possible" is false consensus bias - you're projecting your own feelings onto others who don't necessarily share those feelings, and therefore assuming everyone else must find it boring too.

Anki is also more suitable for beginners than books/shows. Realistically, you can't read a book or watch a movie when you are just starting. Everything will be so incomprehensible that the effort of having to stop to look things up will be overwhelming and tedious. Anki on the other hand can be started from your very first word or sentence.

For me personally, I neither like to dismiss nor focus too much on any one method. I've always learnt best when I put effort into multiple different methods: Anki, books, audio, apps, TV, real life practice, etc. This also helps to keep things fresh and interesting.

I completely agree with your first statement and I try to hand-generate my cards as much as possible.

On the other hand, hand-generation is very time intensive. Having some kind of Anki card for a topic you need to memorise is better than having nothing at all. If LLMs help you write cards that you wouldn't otherwise get around to writing then it can be worth it.

As an example, I've always found Anki really effective for my language learning. But the bottleneck was always finding the time to find good quality sentences from sources like grammar books and then creating the cards. Now I ask ChatGPT to generate me a whole bunch of example sentences for a particular topic or grammar point that I want to master, and I bulk import them into Anki in one go then use AwesomeTTS to create the audio. These cards feel less personal to me because I've lost the benefit of having put in the hard work of creating them myself from source materials. But that's more than made up for by the fact that I'm now progressing through the topics I need to learn at a much faster speed. I'd rather know 1000 words reasonably well than 200 words very well.

After a few repetitions I don't think there's much difference anyway between a card you generated yourself and one you didn't - the SRS algorithm sorts it out for you in the end. The AI generated one might just need a few more reviews/fails/hards to get to the same level of memorisation.

EDIT: I should add that I don't blindly trust ChatGPT's output. My wife is a native speaker for one of the languages, so I always have her check the cards. For my other language, I run the sentences past several other LLM models and I only keep those that all of them agree are correct and idiomatic.

What's the benefit of this over Ansible (which, at a first quick glance, it seems similar in philosophy to)?

Well, there's Guix as an alternative if you want a similar concept but different implementation philosophy. For me the major disadvantage of Guix is lack of package availability compared to Nix.

So your argument is that we do so many terrible things already, that anything else is justified? Surely the better argument is that we should try to stop doing those other things.

> AGI will behave as if it were sentient but will not have consciousness

Citation needed.

We know next to nothing about the nature of consciousness, why it exists, how it's formed, what it is, whether it's even a real thing at all or just an illusion, etc. So we can't possibly say whether or not an AGI will one day be conscious, and any blanket statement on the subject is just pseudoscience.

How can you set a precedent for doing something without doing that thing? Here's a dictionary definition for precedent: "an earlier event or action that is regarded as an example or guide to be considered in subsequent similar circumstances."

But there are dozens of great UXs on top of Git already, so if that's 95% of what jj is about then you seem to be making an argument for jj adding very little value. My understanding is that jj offers more than just UX enhancements.

Then a friendlier and clearer wording for the policy would work better. The current one says you will be "immediately" banned "without recourse" which implies that nothing like you are describing will happen.

> Banning LLMs can result in shorter argument

Can it really? "You submitted LLM-generated contributions" is also highly subjective. Arguably more so since you can't ever really be sure if somethingi s AI generated while with quality issues there are concrete things you can point to (e.g. it the code simply doesn't work, doesn't meet the contributor guidelines, uses obvious anti-patterns etc.).

It's you who isn't reading the policy. What you are reading instead is deleted sections in a Git commit which are not part of the policy.

You've made this comment more than once in this thread. Have you correctly understood that the policy is only the green parts in that link and not the red parts?

Isn't it then more reasonable to have a policy that "people who submit low quality PRs will be banned"? Target the actual problem rather than an unreliable proxy of the problem.

LLM-generated code can be high quality just as human-generated code can be low quality.

Also, having a "no recourse" policy is a bit hostile to your community. There will no doubt be people who get flagged as using LLMs when they didn't and denying them even a chance to defend themselves is harsh.

> How does one show that no assistive technologies below some threshold were used?

In this case, you don't:

> immediately be banned without recourse

In other words, if the maintainer(s) think it's LLM-generated, right or wrong, you're banned.

Interesting. Can you clarify this part?

> there's no reason to have LSD and an adulterant

Why is that? From what I understand LSD is pretty challenging to produce, so isn't there an advantage to an unscrupulous producer using something easier/cheaper?

> If you disregard short lifespans, then intergalactic journeys well below light speed (think 0.1c, 0.2c) is perfectly feasible for such a civilization

They're feasible even with short lifespans with the use of generation ships. Or with suspended animation technology. Given that all three possibilities (life extension, generation ship, suspended animation) are already considered within the realms of possibility by humans (even though we haven't solved any of them yet), it seems a very flawed assumption that no other civilization could solve any of them.

> we probably don't have to worry about anything further than 20 light years away.

On what basis? Are you assuming that malevolent extraterrestrials would be unwilling or unable to travel further than this? Why?

Where did you see a demand? The comment you're replying to merely asked if there is a fork.

I'm curious how people go about finding safe sources for something like LSD (to be clear I'm not asking for anyone to give specific details, just an explanation). While psilocybin is easy to try safely (just travel to a place where it is legal), LSD is illegal in every jurisdiction as far as I know. The thought of ingesting some substance from dubious sources, that could be contaminated with absolutely anything, terrifies me and I honestly don't know how people bring themselves to take this kind of risk. Am I missing the point and there is some safe underground supply chain for this sort of thing?

What on earth are you saying? It's better to kill 20 children than to risk that 2 of them develop peanut allergies? I don't see how this can even begin to be an arguable position to take. And that's ignoring the fact that it isn't even a correct assertion in this case.

> But then they have to remember the username AND the password?

The commenter already acknowledged that the solution has drawbacks. The only claim made was that it solves credential stuffing, not that it doesn't inconvenience the user.

> This would only work for those with a password manager

It would also work for those without a password manager, because they'd have no choice.

> Also, wouldn't this prevent lost password recovery? if you can't identify a user by their email?

They're not mutually exclusive. You can have both. A compulsory unique user ID to login, and an email based password recovery mechanism.

"Unfortunately no single defense provides complete protection"

For the server, maybe. For the user, a password manager (used properly) is a solid single defense solution.

I tend to think it's neither, but rather an inevitable result of the lossy process of condensing legal text (which has been carefully written to include all the nuance the drafter wanted) to something shorter and simpler.

Confirmation bias doesn't really apply to the same extent with law as with other fields.

If the AI says "Fisher v Bell [1961] 1 QB 394 is authority for the principle that an item for display in a shop is an invitation to treat", it's either right or wrong. A lawyer can confirm that it's right by looking up the case in Westlaw (or equivalent database), check that it's still good law (not overridden by a higher court), and then read the case to make sure it's really about invitations to treat.

If the AI says "Smith v Green [2021] EWCA Civ 742 is authority for the principle that consideration is no longer required for a contract in England" it will take you under 1 minute to learn that that case doesn't exist.

Law isn't like, say, diet where there's a whole bunch of contradictory and uncertain information out there (e.g. eggs are good for you vs. eggs are bad for you) and the answer for you is going to be largely opinion based depending on which studies you read/prefer and your biases and personality. It's usually going to be pretty black and white whether the AI was hallucinating. There are rarely going to be a bunch of different contradictory rulings and, in most of the cases where there are, there will be a clear hierarchy where the "wrong" ones have been overridden by the correct ones.

I guess I kind of agree with your disagree, but disagree overall!

The UI in LibreOffice feels quite clunky and outdated and never seems to have been given any thought since the OpenOffice days. But Google Docs is so feature poor that I'd rather live with LibreOffice's UI. Especially as you can adjust to the latter after using the software for a while.

Next step: escape the Google ecosystem. LibreOffice for docs (its far superior to Google Docs anyway). Syncthing for drive.

I'm a legal professional who uses AI to help with my work.

I couldn't ever imagine making a court submission with hallucinated legal references. It seems incredibly obvious to me that you have to check what the AI says. If the AI gives me a case citation, I go into Westlaw and I look up and read the case. Only then do I include it in my submission if it supports my argument.

The majority of the time, AI saves me a lot of work by leading me straight to the legislation or case law that I need. Sometimes it completely makes things up. The only way to know the difference is to check everything.

I'm genuinely amazed that there are lawyers who don't realise this. Even pre-AI it was always drilled into us at law school that you never rely on Google results (e.g. blogs, law firm websites, etc.) as any kind of authoritative source. Even government-published legal guidance is suspect (I have often found it to be subtly wrong when compared to the source material). You can use these things as a starting point to help guide the overall direction of your research, but your final work has to be based on reading the legislation and case law yourself. Anything short of that is a recipe for a professional negligence claim from your client.

Yes, and I agree with that point. I just think you haven't extended it to its natural conclusion which is that if societal structure fails, even physical possession of things you own isn't going to help you if you have gangs of criminals rioting and looting. Consider for example what happened in Haiti in 2018+.

What matters is rule of law, not possession. If you have strong rule of law, you have some protection for whatever you own, possessed or not. If you have zero rule of law, you have no protection for any of it.

The problem is, you're not just writing a large cheque in return for a working treatment. You're writing a large cheque to roll the dice on some research that might (and probably won't, statistically) result in a working treatment. So you need a very large cheque so that you can fund multiple research projects in the hopes that one of them will give you a treatment. Even then, there's the possibility that none of them will work and you've got nothing to show for any of the money spent.