Ages ago, when Chrome was still a new kid on the block, I read an article about it. Turns out, this browser is spying on you so hard that it made me, nope out immediately. Somehow, people missed that article, and others like it, and pretty much everyone started using Chrome on their computer (see also: pro-mobile era).

Especially among 4 year old girls, but the kitty merch is a different story.

A friend advised me to put lots of Hello Kitty stickers on my bike so that nobody would want to steal it. I wonder if the same logic applies here.

This looks like a dream target for hackers. Imagine all the data you could find here. If there ever was a single point of failure, this is it.

Mobile phones and earbuds solve that problem for the most part.

I had very similar experiences around 2015. Before that, it was reasonably easy to use a fully FOSS system, but things have gotten worse over the years. Sure, there’s more FOSS for mobile hardware than ever before, but the world surrounding the phone has moved in the exact opposite direction. Being compatible with the world around you is the problem here.

Back then, I couldn’t find a satisfactory solution. One extreme is to go full on FOSS, and cut at least 50% of the entire world from your life, while the other is to sacrifice your privacy at altar of corporate greed. Between the two there are numerous dissatisfying compromises, and you need to do some soul searching to figure out where you want to draw the line.

Maybe it’s automated propaganda and opinion steering.

Perplexity seems to do a pretty good job, whereas Copilot makes stuff up all the time.

All LLMs I’ve tested had a tendency to agree with my delusions and misconceptions, so you have to be very cautious not to ask loaded questions. If you start misleading the LLM, it will go with the flow and give you a wrong answer.

Copilot and chatGPT prefer to avoid PR disasters, but Mistral has no issues with sensitive topics. Mistral doesn’t really seem to have much opinions about anything so you can dive into any topic you like. The other LLMs do have clear opinions and lines they won’t cross.

Facebook has started using AI bots to boost engagement (and revenue). Is Reddit trying to do the same thing?

It works! I typed in “blockbuster” adjusted upload time filter to “today” and started scrolling. Just scroll past anything that is less than 1 h long, and you’ll find movies pretty easily.

Before long, I found a few movies that had very special names like: “Rotten Tomatoes Full Movie | Blockbuster Movie 2025 | Action Movies Hollywood 2025 in English 2160p” or “60’s Sci-fi | The Day Mars Invaded Earth | Full Movie”. I also tried a bunch of other related keywords like: “movie”, “hollywood” and “scifi”. Surprisingly, finding bollywood movies is very easy.

The only problem is that you’ll never know what you’ll find. If you want to watch jurassic park, you could search for “dinosaur movie”, but I couldn’t make that work. Searching by genre seems to work reasonably well, but you need to get creative with the keywords.

When you do find a movie, copy the name into a text file and start collecting these silly names. Pretty soon you’ll notice which words appear frequently, so start searching for those. There’s no single universal word that always works. Here’s are a few words that seems to work for me: Rotten Tomatoes, Full Movie, film, blockbuster, action movie, hollywood, 2160p, adventure, Thriller, mystery movie, free movie

Thanks for the detailed explanation.

Close the door, turn off the lights, wear earbuds, pull a duvet over your head. Works every time. Might look a bit suspicious, but at least it’s private.

Seriously though, it’s getting pretty bad. I’m currently shifting my video watching habits away from YT. I have a feeling that sooner rather than later I may have to quit YT completely.

If you’ve never thought about threat models, here are some questions to get you started.

1. List the assets that require protection. (Like credit card info)
2. Who might want to gain access to those assets? (e.g. Hackers)
3. How can you mitigate the risks? (Updates)

You may have a lot to say, so write things down to clarify your thoughts. Once your threat model document is complete, it will be easier for you to figure it which tools you really need, and which ones are only nice to have.

That’s a fun problem, if you enjoy recreational mathematics. However, if you need to deal with this nonsense on a daily basis, it get’s old pretty fast.

Oh, yeah, the pounds, shillings and pence system was just epic. One pound (money, not weight) is 20 shillings (because why not), and one shilling is 12 pence (just for maximum confusion). In order to use measurements or money, you need to memorize a bunch of arbitrary conversion factors, which was obviously loved by everyone.

Using a 100% de-googled Android would be best, but it comes with some serious compromise, so this option isn’t for everyone. The second best options are also pretty good and quite practical in comparison.

It really depends on your situation and the kinds of threats you’re trying to mitigate. If you need to have a spyTV, spyPhone, a spyPad or something similar in your life, using DNS filtering may be an appealing option. Ideally, you would have zero iOS or Android devices in your household, but life isn’t ideal.

How about a NextDNS subscription? You can use that to filter out a bunch of creepy tracking, ads and other stuff. Ideally, you would only use devices and software that don’t do any creepy stuff, but the world is what it is, so using a custom DNS is a good idea.

I’m not touching an app owned by Meta. If people want to message me, they can use Signal.

Turns out, this policy has weeded out lots of useless conversations from my life. So, I see it as a win.

Thanks. That was a good summary, and I appreciate that you brought up threat models.

People should think about what kinds of threats are worth their time and money. If that list of threats contains something where a VPN can help, you should totally consider getting a VPN. If your threat model doesn’t include things like that, VPN might not be the solution you’re looking for.

Brining up trust was another good point. People should think about how much they trust their ISP or some VPN company. Obviously, you can’t trust every VPN company out there, but where you draw the line is closely connected to your threat model. For example, if you are a journalist in dangerous country, picking the right company is a matter of life and death. If you are in a safer environment, your threat model is probably very chill by comparison, so you might be fine with some less secure options.