Yep. AGI is still science fiction. Anyone telling you otherwise is probably just trying to fool investors. Ignore anyone who is less than three degrees of separation away from a marketing department.

The low-hanging fruit is quickly getting picked, so we’re bound to see a slowdown in advancement. And that’s a good thing. We don’t really need better language models at this point; we need better applications that use them.

The limiting factor is not so much hardware as it is our knowledge and competence in software architecture. As a historical example, 10 short years ago, computers were nowhere near top-level at Go. Then DeepMind developed AlphaGo, which was a huge leap forward and could beat a top pro. It ran on a supercomputer cluster. Thanks to the research breakthroughs around AlphaGo, within a few years had similar AI that could run on any smartphone and could beat any human player. It’s not because consumer hardware got that much faster; it’s because we learned how to make better software. Modern Go engines are a fraction of the size of AlphaGo, and generate similar or better quality results with a tiny fraction of the operations. And it seems like we’re pretty close to the limit now. A supercomputer can’t play all that much better than my laptop.

Similarly, a few years ago something like ChatGPT 3 needed a supercomputer. Now you can run a model with similar performance on a high-end phone, or a low-end laptop. Again, it’s not because hardware has improved; the difference is the software. My current laptop (2021 model) is older than ChatGPT 3 (publicly launched in 2022) and it can easily run superior models.

But the returns inevitably diminish. There’s a limit somewhere. It’s hard to say exactly where, but entropy’s gonna getcha sooner or later. You simply cannot fit more than 16GB of information in a 16GB model; you can only inch closer to that theoretical limit, and specialize into smaller scopes. At some point the world will realize that trying to encode everything into a model is a dumb idea. We already have better tools for that.

I don’t have this chart in my review, I guess because I was 100% Linux. Kind of surprised I didn’t have a little Mac time in there, since I do sometimes use Steam on Mac while traveling. But thinking back, I guess it’s been a while.

I know this is from 2015, but even then, it was a bit late to make this argument. This was already mainstream enough in the 90s to be the punchline in syndicated comic strips. By 2015, we already had “customer experience engineers” (i.e. tier-1 helpdesk). The ship has not only sailed, it has sunk.

Anyway, the phrase originated in an era when programming was very different from what it is today, when most programmers came from a background in electrical engineering or something along those lines.

Apple’s monitors have an entire OS in them. They have much of the same internals as an iPad. Honestly, I have no idea why, because they don’t do anything especially fancy.

Samsung makes “smart monitors” with Tizen or some shit like that.

DBAs think everything is a database.

Perhaps it would be better to link directly to the more scientific sources linked in the article:

https://scribe.rip/cuinfoscience/an-exploration-of-the-twitter-to-mastodon-migration-21c15c4336f2

https://doi.org/10.1145/3392847

The author of the article contributed to or co-authored those as well. The article is very general writeup, but there is real science here.

This will have no effect on torrenting or other P2P protocols. Your IP address will still be out there.

I’ve never heard of DMCA warnings based on DNS requests. That doesn’t really make sense.

Yeah, Matrix is a very, very hard sell. I mean, “normal” people (for lack of a better term) are put off by Mastodon, and Matrix is a hundred times more complicated to join. I’m also not sure what it would look like to use Matrix the way I use Discord. Perhaps there is functionality in Element/Matrix I have never explored since I use it more for messaging and group chat, not for communities with multiple channels like IRC/Discord/Slack.

In any case, Discord is too entrenched to be replaced by something that is merely technically superior, or even more user-friendly. Realistically, you can’t migrate entire communities if they’re bigger than a tight-knit IRL friend group, and even that is hard. That seems to be the only reason X still exists.

Almost, yeah. Certainly the big corps.

This is why I strongly favor services that use end-to-end encryption or do not store history in the first place.

There are not many times when I’ve needed to search back through history on a Discord server, and every time I have I thought to myself “this would be much better on any platform besides Discord”. Discord would, IMHO, be a better product if they did not retain history forever.

Ditto for Slack. Slack has the additional gall to limit access to that data unless you pay for a premium plan, despite the fact that they keep the data forever regardless (as evidenced by their occasional free trials which magically bring all history back, and some search tricks you can use to access old posts regardless).

Both Slack and Discord have lulled their user base into a false sense of privacy. Nothing you post there should be considered private.

It doesn’t really matter if they do or don’t. What matters is that they can change their TOS at any time, they keep an archive of all historical data, and you will have pretty much no recourse no matter what they decide to do with it in the future.

Who knows what will happen to Discord in five or ten years?

They might get bought by a narcissistic billionaire.

They might sell all their data to Google for training AI.

They might go bankrupt and sell off their assets to the highest bidder.

They might have an IPO and begin the usual value extraction at the expense of their users.

I know, I know…crazy ideas, right? When has anything like that ever happened?!

I try not to judge people…unless I see them right-clicking to copy and paste. Ew.

My guess is that this is a teenager, and this is probably their first experience with git and version control in general. Just a hunch.

Anyway, it is reasonable to expect a mainstream GUI app from one of the largest companies in the world to be approachable for people who do not know all the inner workings of the command line tools that are used behind the scenes. And it is reasonable to expect any destructive action to have clear and bold warnings. “Changes will be discarded” is not clear. What changes? From the user’s perspective, the only changes were regarding version control, so “discarding” that should leave them where they started — with their files intact but not in version control.

Have mercy on the poor noobs. We were all there once.

I feel bad for this kid. That really is a bad warning dialog. Nowhere does it say it’s going to delete files. Anyone who thinks that’s good design needs a break.

Half the replies are basically “This should be obvious if your past five years of life experience is similar to mine, and if it isn’t then get fucked.” Just adding insult to injury.

This is good advice, because email is very difficult to make reliably private. However, it’s not the best you can get. Tutanota, for example, stores headers with E2EE, and still has a search function.

The goal should be to make it as private as it can realistically be. Ideally, any cloud service you use should only store end-to-end encrypted data.

I’m not trying to shit on Proton — it’s a huge step up from the popular mainstream email services, and the inclusion of cloud storage makes it a much easier transition than going piecemeal with 2-5 different services.

Not the encrypted mail, mind you, because they can’t do that

Just want to point out for anyone new that ProtonMail does not use E2EE for email headers. That means they CAN access your subject lines, to/from fields, and other email headers. That means they CAN be forced to hand it over to the government.

Source: https://proton.me/support/proton-mail-encryption-explained

Subject lines and recipient/sender email addresses are encrypted but not end-to-end encrypted.

Personally I am disappointed in a lot of Proton’s wording about this. They frequently promise they can’t access “your data” and “your messages” when they do, in fact, store potentially sensitive data in a format they CAN access.

Using an ad-blocking DNS server solves most of those problems.

You should still use a browser extension on top of that for pattern-based URL blocking, but a DNS-based blocker should be your first line of defense.

A good ad-blocker goes a long way. You can block all Google domains with minimal impact to non-Google services.

30 years ago, maybe. Post-Napster, not relevant. Most online piracy is non-commercial now, and it’s still illegal across most of the world.

In practice, Python is not easy to learn programming with. Not at all. I see beginners wrestling with Anaconda and Jupyter notebooks and I weep.

The fact that pip is intentionally broken on macOS and some modern Linux distros sure doesn’t help. Everything about environment management is insane.

This is how Arc behaves by default, and it drove me crazy. The cons of it appearing when I don’t want it far outweigh the pros of saving me a single click when I do want it.