if you want to take OpenAI’s own research into account
No thank you.
OlympicArena validation set (text-only)
“Our extensive evaluations reveal that even advanced models like GPT-4o only achieve a 39.97% overall accuracy (28.67% for mathematics and 29.71% for physics)”
Unlike this year when LLMs are more of a huge scam.
Oh Jesus. Really?
Holy crap. That explains nearly everything. The only things that still seem weird are:
Still, though, the idea that it’s not “remembering me” and probably is just giving people that timestamp when they search that term by default even if they’ve never run across that video before seems like the most likely explanation.
Oh, and I did take a minute to go try this on (a fairly outdated version of) Firefox on another Arch Linux laptop on which I wasn’t logged in and all my cookies/history/form data/etc had all been deleted immediately before. I did get the indicator on that video when searching “gnu taler”. Which definitely seems like more validation of this theory.
Thank you for your input!
While logged out, https://www.youtube.com/feed/history gives me the following:
And it’s still showing the indicator on the “gnu taler” search results page.
I suppose it might be worth closing my browser, opening my browser, going to YouTube, logging in, and checking that page, though. It might at least give some information or something. I’ll try that here and see if it lists the video in question. I’ll update when I’m done.
Edit: That video about GNU Taler does not show up in my viewing history while logged in. I tried viewing a random video while logged in and checking my viewing history and that random video shows up. But not the GNU Taler one that still has the indicator. I’m starting to think I’m losing my mind. Lol.
Right, the link has a timestamp in it. But I don’t know how the search results page knew what timestamp to put in the link to the video even after switching to TBB.
Yeah, forgot to mention I hadn’t logged in or anything. But that’s right. I didn’t log in during any of that testing. I’ve edited that detail in.
Still weird that it could possibly fingerprint even while using TBB, though.
Not sure I understand what you’re getting at here.
Yes, I linked to the video and didn’t think to remove the t=273s bit when I included the link in the OP. And, yes, I understand that having a &t=273s in the url makes it start not right at the beginning. My question is how did it know where to start (and how much red bar to show on the video thumbnail in the search results) given that my cookies had been deleted and, on subsequent tests, I even switched browsers.
I was purposefully telling my browsers to forget all the information YouTube could use to remember that and it still remembered somehow.
Now, I am concerned regarding the privacy aspect of how on earth it still persisted in TBB. But even when sites fingerprint you, if you delete your cookies they almost always at least pretend not to know you when you visit. I’d expect YouTube/Google to use fingerprinting to sell my information and do targeted advertising or whatever. But it’s weird that they’d even let on to me that they had figured out who I was even though I wasn’t sending them any cookies.
Also, did you return to that video with the same IP address as when you first watched it?
That’s (part of) why I tried Tor Browser Bundle, though. Because it would give me a different IP address. (And when I visited YouTube via TBB, it gave me the little superscript after the YouTube logo indicating a different country than I was in.)
I’ll just assume you didn’t log in to youtube when watching. :)
Ha! Should have thought to mention that. But yes, you’re right. I didn’t log in or anything. (And for that matter, in every test I did, when I first got to the home page, I got the “search to get started” prompt that YouTube gives as of pretty recently when you don’t have any cookies on visiting the index page.)
Oh! That’s good to hear. Honestly, that issue has kindof pissed me off enough at Rossmann specifically that I kindof quit watching his YouTube videos and stuff. So I very much haven’t been following him or FUTO.
I wonder if the FUTO website still claims that they require all projects to be or have a plan to become specifically “Open Source”.
Edit: Yup. They still say “All FUTO-funded projects are expected to be open-source or develop a plan to eventually become so” on this page. Maybe that means that they intend for Grayjay to “develop a plan to eventually become” properly Open Source and not just “source first”.
Just in case this matters to OP or anyone else in this thread, Grayjay isn’t Open Source, despite Rossman’s and FUTO’s claims to the contrary. Its license disallows any commercial use of Grayjay, and also disallows removing any features related to paying FUTO. Which disqualify Grayjay as “Open Source” by the OSI’s definition.
And consequently, F-Droid won’t distribute Grayjay unless they change their license.
A lot of user fingerprinting techniques rely on JS. Plus, by shutting off JS, you reduce the attack surface of your browser. If, let’s say, there was a zero-day vulnerability in Firefox that required JS to exploit, you’d be shutting off that whole means of attack if you blocked all/most JS out there on the internet. Mining cryptocurrencies on your computer via your browser can only be accomplished with the help of Javascript. A lot of forever cookie techniques require Javascript.
uBlock origin is for kindof a different use case. It’s for if you’re on one website that you don’t necessarily suspect of evil dealings that might include buttons (like social media sharing buttons, for instance) or other scripts (like ad displaying scripts or analytics scripts) from third parties that might include evil tracking stuff. If I started a blog on https://theawesomeestblog.com/ and included script from Facebook that puts a share button on my page, and if you then visited my blog, Facebook would know because your browser would make requests from your IP with cookies they’d placed on your brower previously and JS included with the button could very well be used to do additional fingerprinting.
NoScript is for (among other things) when you don’t even necessarily trust the website you’re purposefully visiting. Like, I don’t know if cnn.com mines Bitcoin via JS on users’ browsers (and, honestly, it seems a little unlikely to me, I think), but if I disallow JS on cnn.com, then when I click a link in Lemmy to a cnn.com article (and maybe I don’t even really know I’m going to cnn.com when I click the link – it might use a link shortener or something – or maybe it’s not cnn.com, but some reasonably-trustworthy-sounding news-y-sounding domain that I haven’t heard of before), I know it’s not mining Bitcoin on my machine.
Oh, and as others have said, NoScript is Open Source. Says so right near the top of the home page.
/c/Libertarian is that way:
/c/Cryptocurrency is over there:
Thank you for bringing more awareness of this. I’m what you might call an “AI skeptic” and don’t really care what happens in the AI space as long as it doesn’t screw up things I care about.
But I care deeply about FOSS and AI is screwing it up. I don’t want to have to explain why XYZ thing absolutely is not Open Source and that “Open Source” has a specific meaning beyond “you can look at (at least some of) the source code.”
(Compare it to the term “hacker” that has among at least a lot of muggles taken on the exclusive meaning of committing some kind of fraud with computers. Originally it meant something very different. And it’s unfortunate the world has forgotten the old meaning.)
Another project that is diluting the term “Open Source” is Grayjay, a video streaming app that is a FUTO project (and FUTO is a Louis Rossman thing.) Rossman has called it Open Source in YouTube videos, but it’s not Open Source. (The license is here and forbids things like “commercial use” (selling the software or derivative works) and removing facilites for paying the FUTO project from derivative works. Which is a lot less restrictive than the license was last time I checked it. Previously it didn’t allow redistribution or derivative works at all. But it’s not Open Source even now.)
Minetest is an excellent candidate for this use case, I’d say. I play Minetest on a Raspberry Pi 4 connected to a Minetest server hosted on a different Raspberry Pi 4 (both running Arch Linux Arm) over Wifi regularly and don’t find latency or lag or anything to be a significant issue.
The world I’ve been playing in that particular way is on the game VoxelLibre (Formerly Mineclone 2). It’s intended to mirror Minecraft’s functionality as closely as possible. It’s not 100% implemented, but a lot of it is implemented, and it’s very playable. The default Minetest Game is great too. It lacks mobs, but mobs can be added with mods like the various Mobs Redo API plugins. (And it’s easy to have mobs but not enemy mobs that can kill the players if that’s the vibe you’re going for.)
Generally, I’d recommend F-Droid as a best first place to go for app recommendations if you’re interested in privacy and such. The vast majority of software there will be ridiculously in line with your privacy wishes, and the ones that have minor caveats will be explicitly marked as such. And its search interface is very decent.
Not saying I disagree, but what specifically would you prefer? Gecko? Or is there something better? (Maybe I’m just engaging in wishful thinking asking if there’s something better.)
Regardless of 4chan’s privacy situation, it’s a total cesspool. Nothing worthwhile happens on 4chan.
Only thing that’s guaranteed is that you’ll never actually get to the sexy goblin.
The README in the repo indicates it’s based on the NEO-PI, which is kindof the gold standard in personality tests at least right now from what I understand.
Book recommendation for folks who might want to know more about the topic of personality psychology. Me, Myself, and Us: The Science of Personality and the Art of Well-Being by Dr. Brian Little.