Hearing “not particularly prone to faults” in context of a nuclear powerplant isn’t exactly comforting :D

True, but iirc there are several alternatives, from different repositories, and i was unlucky enough that j choose the wrong one for the first time.

Exactly! AdNauseam is my inspiration, unfortunately I dont use it right now because it doesnt work with my setup, due to Mullavad in combination with PiHole being too good at their job, so it cant really click on the adds even if it wanted to.

It would ideally require a solution that uses its own DNS server, so you can keep on using PiHole, and prefferably avoids living in your browser so you can keep it locked down for privacy.

One of the projects I have in mind is to explore some kind of “offensive privacy”, where the focus would not be on not being trackable, but on your computer spewing random bullshit and behavior into the algorithm to confuse it, and have it learning on behavior that’s not really true, but only generated. This will enable you to kind of fight back and if done by enough users even reduce the effectivness of ML algorithms, since they would be learning bullshit. Unfortunately, the scale required to effectively affect the learning process of ML models would be enormous, so it’s not really feasible, but I think it’s still better than just “staying hidden”.

With the advances in AI, creating a tool like that, that would simulate several random user behaviors on your IP/fingerprint, shouldn’t really be that hard.

And as an added bonus - if it clicks on adverts, it’s costing someone money. Fuck corporations.

I solved that by social engineering our IT to join my “Windows” computer into the domain, which was actually just a Windows VM. They didn’t notice, and I’m free to Linux away.

I’m also running NVIDIA (RTX 4070), and while I did have to try drivers from a few different sources, I eventually got it working pretty quickly.

But my mistake was choosing an OS that doesn’t bundle non-free drivers (Fedora), from what I’ve heard some distros like Ubuntu come with NVIDIA support by default, so I guess that’s also an option.

I literally did this two weeks ago, switched Win11 for Fedora and so far it has been an amazing experience. So far, I only had to dual boot to Win once, and that was because I wanted to play some SteamVR games, which is the only thing I didn’t manage to get working (I know there’s ALVR, but SteamVR refuses to launch for me unfortunately).

Just go for it, get a new SSD drive and dual boot your choice of distro. You can always go back, and unless you use bitlocker you can just access your windows files from the Linux, so there’s not need to move stuff around that much. With dualboot, you have nothing to loose.

The biggest problem i have with my data being collected, analyzed and used is in the fact that it will almost certainly be used to teach a ML model about how to better manipulate with people like me - the people that are privacy conscious and are trying as much as possible to reduce their fingerprint.

That data is invaluable, and if there does exist a way how to target even people like that, which there probably does since we’re only humans after all, the ML model will eventually figure it out. And they have literally billions of people to experiment and learn on.

Now, we already know from a few leaked studies made by Facebook that they cab already pretty well manipulate people into mostly whatever they choose. Take a hypothetical situation where you get a crazy out-of-touch billionaire, who decides to buy a large social network company, and then decides “Hey, I really want this candidate to win. Tune up the algorithms!”.

And the ML models will get a clear goal, that has been already proven to just work pretty well at influencing user behavior. And any data you give them, it helps the model to fine tune into influencing people like you . Which would also be really hard to prove, because ML models are by definition black boxes that are really hard to reverse engineer, and proving that it was trained to do this is AFAIK almost impossible.

I don’t want no part in that. Thankfully, all the large social networks have CEOs that are reasonable and would never try something like that, right?

And one more thing - you may not think that data about your behavior are of interest to anyone right now. But look at China and their Social Credit. And imagine how would have I.e holocaust turned out, if the government had access to all the data, opinions and profiles of people that are being collected now.

Oh, you mentioned you sympathize with the Jews three years ago in a private message? Well, let’s hope the country you live in never ends up in a situation where that could be a huge problem for you or your family.

So, every time any site is offering a “personalized, curated list” for you (I.e the google search result, or YouTube recommended videos), assume you are potentionally being manipulated, and avoid the site altogether- because there’s no other way how to prevent it. The ML model knows that you know, and is already trying to figure out how to manipulate people that are taking care not to be. And if there is a way, it will figure it out with some success.

Yes, because you don’t want to provide them your content for free, so they can continue building engagement on your behalf.

I’ve recently switched to Kagi and it has been an amazing experience so far. I definitely recommend that to anyone who can spare the 5-10$ a month. I like their business model, and the way I can customise results to I.e always ignore reddit posts, while still maintaining privacy because they are not in the ads business (yet?).

And the results are usually pretty on spot, while also avoiding the major ai/spam blog posts by default.

I guess it may be a combination. I use Mullvad VPN + Mullvad browser. Maybe they only show it if the can’t fingerprint you enough? I.e by a combination of IP, and your browser + extensions + tracking cookies fingerprint?

Yup, it’s a VPN. If I turn it off, it works properly even in the same browser (Mullvad). But I refuse to cave in, so if they every start checking for VPNs even in the old Reddit, or disable it altogehter (which I’m sure will be soon, since you can also use it to bypass “Log in to see NSFW content”), I’m out for good, and will finally block Reddit at my Kagi.

And every time a Reddit results show up, I’m immediately reminded why I don’t want to go there by an error telling me that I can’t use the site without logging in.

Fortunately, just changing the link to old.reddit.com still works even through VPN, but fuck this behavior. I do that only for questions I really need an answer and couldn’t find anywhere else, and most of the time the replies are shit anyway.

Thank you for this, I’ve never heard about this show, watched the first pilot, and am hooked. It looks amazing!

The whole article felt like bullshit, and if I went to college and had to sit through listening to something like that - just random doomsaying without any proof, that doesn’t even makes sense (“we won’t need real CS engineers, but we will need prompt engineers to somehow convince the model and discover the correct prompt, since we don’t know why it’s replying as it is” - oor, maybe you can just leave out the AI step and have a real person do it.), and it’s lacking any sensible point.

Oh, right. I get it now. So that’s the point.

It happens partly through the use of platforms like Fixie, his company’s platform for easily creating AI-based applications

I would be so mad if something like this happened on my college.

You are right I shouldn’t have equaled bitcoin with the rest of the crypto ecosystem. While most crypto is utter scam, it’s true that there have been some slight advances here and there, and there are coins that may be actually useful for some cases, mostly Monero and I suppose Ethereum. I’d still say that crypto has done more harm than good in the world, and I say that as someone who’s really focused at privacy, care about it a lot and have invested significant amount of time and effort into staying as private as possible.

But it’s great that Ethereum managed to solve most of the issues with Bitcoin - unless I’m mistaken, it’s not really used for investment speculation, and if it managed to keep the energy requirements low, that’s good. But last time I remember researching about blockchain (it was few months, so feel free to correct me), isn’t it running into serious issues with ledger size, that makes it infeasible for long-term (decades) of use, without sacrificing some of it’s guarantees? Which is one of the main issues with blockchain tech in general, that I don’t think has been solved so far.

I just hope bitcoin will finally die. It’s literally just wasting absurd amount of energy, only to allow scammers to scam billions of dollars from victims, and regular people to steal from eachother by investing into it. I mean, if the only use of bitcoin by now is for speculation and investment, then it means that any dollar you made, you literally stole from someone else who will be left with useless bitcoin once it’s all over. There’s no value, and with the ledger getting bigger and bigger, and bitcoin more expensive to mine, it will eventually be worthless. And we all know it, so anyone who makes thousands of dollars, there’s someone who probably financially ruined himself by making a wrong and stupid investment at the wrong time.

I hate crypto so much :D.

700kWh per transaction? That’s absurd amount of power. That’s 70 EUR of energy per one transaction at current (EU) exchange price.

Is there anyone here knowledgeable enough about this issue to say whether those numbers are correct, or just an overestimate? It feels wrong.

After several of my favorite songs disappeared from Spotify, I’ve adopted a different approach to music.

If I see on on a band show merch stand, I buy a cassette. It’s more of a novelty item and a way to slightly support the band. While I do have a portable tape player, I only rarely take it out. I switched from LPs to tapes because of the costs and huge effort associated with playing or storing them (that is, if you do it right are are not OK with fucking up your LPs), but tapes are cool and don’t have that many storage or playing problems.

Other than that, I’ve stopped paying for any kind of streaming services, and save the 10$ per month to just buy one or two (new or old) albums from my favourite artists on Bandcamp, that I’ve spend the last month listening to the most. The albums I buy I add to my NAS library, which usually replaces stolen copies of said albums that I’ve previously got from Redacted.

This allows me to keep a pretty expansive library, by just stealing what I need, but with a promise that I’ll eventually buy the album (using the money I saved on streaming services), if it’s something that I’ve listened to extensively. I’m also not at mercy of streaming services, that can take away my music whenever they decide to.

So far I’ve been doing this for a few years, and even increased my budget for just buying albums if I can’t immediately find them on Redacted.

That is true, but can’t they (a company that wants to, not the goverment) do that already if they want to, under ChatControl 1.0? And I wouldn’t say that whether a service is E2EE or not makes any difference here - scanning private user messages shouldn’t be allowed, whether they are encrypted or not. IMO if ChatControl 2.0 passed and was made mantadory for everyone, the fact that it is mostly noticable on E2EE apps is only a side-effect of blanket surveilance, and not the main issue with the proposition.

What’s the point of them agreeing that they will let the 1% of users of E2EE services keep their privacy, while they already scan 90% of communication (I mean, just GMail + FB/IG + iCloud, that is already being scanned, makes for most of the worlds communication) for the past year or so?

Now I’m curious whether Facebook/Instagram, who does offer encrypted chats and also scans all your content under ChatControl 1.0 voluntarily, also scans the encrypted chats or not. I’d vager they do, but that’s just a speculation.

But they did briefly mention that they will begin “phasing out” chatcontrol 1.0. I wonder what does that means, and how long will it take.