This is good advice for all tertiary sources such as encyclopedias, which are designed to introduce readers to a topic, not to be the final point of reference. Wikipedia, like other encyclopedias, provides overviews of a topic and indicates sources of more extensive information.

The whole paragraph is kinda FUD except for this. Normal research practice is to (get ready for a shock) do research and not just copy a high-level summary of what other people have done. If your professors were saying, “don’t cite encyclopedias, which includes Wikipedia” then that’s fine. But my experience was that Wikipedia was specifically called out as being especially unreliable and that’s just nonsense.

I personally use ChatGPT like I would Wikipedia

Eesh. The value of a tertiary source is that it cites the secondary sources (which cite the primary). If you strip that out, how’s it different from “some guy told me…”? I think your professors did a bad job of teaching you about how to read sources. Maybe because they didn’t know themselves. :-(

I think it was. When I think of Wikipedia, I’m thinking about how it was in ~2005 (20 years ago) and it was a pretty solid encyclopedia then.

There were (and still are) some articles that are very thin. And some that have errors. Both of these things are true of non-wiki encyclopedias. When I’ve seen a poorly-written article, it’s usually on a subject that a standard encyclopedia wouldn’t even cover. So I feel like that was still a giant win for Wikipedia.

I think the academic advice about Wikipedia was sadly mistaken. It’s true that Wikipedia contains errors, but so do other sources. The problem was that it was a new thing and the idea that someone could vandalize a page startled people. It turns out, though, that Wikipedia has pretty good controls for this over a reasonable time-window. And there’s a history of edits. And most pages are accurate and free from vandalism.

Just as you should not uncritically read any of your other sources, you shouldn’t uncritically read Wikipedia as a source. But if you are going to uncritically read, Wikipedia’s far from the worst thing to blindly trust.

I don’t think the article summarizes the research paper well. The researchers gave the AI models simple-but-large (which they confusingly called “complex”) puzzles. Like Towers of Hanoi but with 25 discs.

The solution to these puzzles is nothing but patterns. You can write code that will solve the Tower puzzle for any size n and the whole program is less than a screen.

The problem the researchers see is that on these long, pattern-based solutions, the models follow a bad path and then just give up long before they hit their limit on tokens. The researchers don’t have an answer for why this is, but they suspect that the reasoning doesn’t scale.

Just saw: they’re changing it back because you don’t appreciate it enough :(

Thanks for linking that. Reading the paper, it looks like the majority of the “self-host” population they’re capturing is people who have a WordPress site. By my reading, the wording of the paper would disqualify a wordpress.com-hosted site as “self-hosted”. But I’d be very suspicious of their methodology and would expect that quite a few people who use WP-hosted reported as self-hosted because the language is pretty confusing.

I was at Google when they announced that only AI-related projects would be able to request increased budget. I don’t know if they’re still doing that specifically, but I’m sure they are still massively incentivizing teams to slap an “AI Inside” sticker on everything.

To someone watching network traffic, a VPN connection looks like two machines exchanging encrypted packets. You can’t see the actual data inside the packet, but you can see all the metadata (who it’s addressed to, how big it is, whether its TCP or UDP, when it’s sent). From the metadata, you can make guesses about the content and VPN would be pretty easy to guess.

When sending a packet over the Internet, there’s two parts of the address: the IP address and the port. The IP address is a specific Internet location, blocks of IP addresses are owned by groups (who owns what is public info) and there are many services that do geo-ip mappings. So if you’re connecting to an IP address that belongs to a known VPN provider, that’s easy.

The second part of the address is the port-number. Servers choose port-numbers to listen to and the common convention is to use well-known ports. So, for example, HTTPS traffic is on port 443. If you see a computer making a lot of requests to port 443, even though the traffic is encrypted we can guess that they’re browsing the web. Wikipedia has a list (which is incomplete because new software can be written at any time and make up a new port that it prefers) and you can see lots of VPN software on there. If you’re connecting to a port that’s known to be used by VPN software, we can guess that you’re using VPN software.

Once you’re running VPN software on an unknown machine and have configured it to use a non-standard port, it’s a bit harder to tell what’s happening, but it’s still possible to make a pretty confident guess. Some VPN setups use “split-tunnel” where some traffic goes over VPN and some over the public Internet. (This is most common in corporate use where private company traffic goes in the tunnel, but browsing Lemmy would go over public.) Sometimes, DNS doesn’t go through the VPN which is a big give-away: you looked up “foo.com” and sent traffic to 172.67.137.159. Then you looked up “bar.org” and sent traffic to the same 172.67.137.159. Odds are that thing is a VPN (or other proxy).

Finally, you can just look at more complex patterns in the traffic. If you’re interested, you could install Wireshark or just run tcpdump and watch your own network traffic. Basic web-browsing is very visible: you send a small request (“HTTP GET /index.html”) and you get a much bigger response back. Then you send a flurry of smaller requests for all the page elements and get a bunch of bigger responses. Then there’s a huuuuge pause. Different protocols will have different shapes (a MOBA game would probably show more even traffic back-and-forth).

You wouldn’t be able to be absolutely confident with this, but over enough time and people you can get very close. Or you can just be a bit aggressive and incorrectly mark things as VPNs.

I have no idea how well it works in reality, but I can imagine the Lifetime Pass being a good business model for them: only the most enthusiastic user will pay for 3 years up front (lifetime currently costs 3x the yearly). So when they get a Lifetime pass they’re getting 3 years paid up front and an evangelist who will probably tell their friends about Plex. If that Lifetime subscriber gets even one person to sign up for a yearly sub who otherwise wouldn’t have, then Plex came out ahead.

Sure, I’m not saying Plex has to do a single-payment model. Just that it’s a think that’s been done successfully (and for longer than Plex has existed). Everyone’s pushing subscription models so hard that it’s easy to think “this is the only possible way that anything can work”.

I like my Shield TV: https://www.nvidia.com/en-us/shield/shield-tv/

I did need to install a custom launcher on it when the standard AndroidTV launcher added ads.

Lots of businesses have and do exist without a subscription model. I’m fond of the Paprika Recipe Manager, for example, which asks a one-time payment for each major version. All commercial software worked this way in the 80s.

It’d be way less offensive if it was just present as an option, instead of dancing around flashing at me