[mikewarot]

Enshittification is caused by the lack of secure general purpose computing, and will continue until we get it back.

We had it back in the 1980s, when we booted from write protectable floppy disks, and could easily verify backups. What nobody realized at the time was that was a capability based security system. Very coarsely grained, but you couldn't accidentally fubar everything. It was easy to know what was being risked at any given time.

We used "shareware" disks, typed in things from magazines, and generally could run ANYTHING, in almost perfect safety. We always had our known good boot disks, and their backups, and our multiple copies of our data.

Until we can run any old random executable, we're going to avoid "untrusted" sites, and stick to the walled gardens our friends hang out at. Capabilities based systems like Genode might be able to get us there.

VMs and Containers were a make-do / very crude version of capabilities... you specified what disks a virtual machine could access, thus it was better than nothing.

WASM is the latest attempt down this road, and I only hope they don't "improve" it by destroying the capabilities model it provides.

What's really sad is that this was all solved back in the 1970s, and it's only the accident of history that we don't have capabilities based systems available that a normal person could use. In response to the need for a single computer to handle multiple levels of classification due to the needs of the Air Force in the Viet Nam conflict, research was done, and solutions were found.

---

Think of the difference between handing someone $5, or your credit card.... which is safer? You can only lose $5 in the first case. That is capability based security.

PS: Sorry, I write defensively because of the way replies tend to work here on HN.

[rlupi]

> What nobody realized at the time was that was a capability based security system.

that's a very interesting idea.

[candiodari]

But this ignores that these solutions need centralized capacity. Games or apps that work this way still work, it's not a problem and they can be isolated. It's web browsers.

The problem is that "communities" need a way to store the messages on them. You can try decentralized, but then people need to provide (and support) a centralized database (or at least disk capacity)

How could you provide that in a centralized way, given that always-on machines mostly don't exist anymore.

[AstralStorm]

A decentralized database is feasible but initial sync to find nodes will take many minutes. Typically those get accelerated by keeping a short centralized list of known potent nodes.

The problem really is not that, Freenet nodes store much more data than socials have in total, the problem is that multitude of services means multitude of filtering and means of discoverability without overloading users. That is an extremely hard problem, not everyone wants to be a moderator, distributed reputation systems are hard to set up and defend against attacks while still keeping any degree of privacy. You cannot depend on user supplied tags without reputation, you cannot know if it's posted to the right space, you cannot get reputation without someone sitting around and giving it out. (confirming it)

Nasty people will give negative or positive reputation for wrong reason, a single number is insufficient to express why you believe something is good or bad, a multitude of numbers dilutes it.

You see it on twitter and reddit with people liking something only because of the poster. Or posting controversial garbage just to get eyes on their other things. And politics gets everywhere.

Plus you really cannot monetize it, so greedy people and companies will skip it besides spamming. (Spam is a solved problem in a decentralized way, we do have three nines bayesian spam filters.)

I recommend reading the history of the original federated newsgroups. They even still are around and slightly active. There have been board splits over moderation, issues with spam and terrible content etc.

[mikewarot]

>But this ignores that these solutions need centralized capacity.

No, I know there has to be a server somewhere, but you have to have your own secure computing system, in order to be able to just try out new things freely.... we used to have this back in the 1980s.... we need to get it back, otherwise we're not free to experiment with new stuff, and as a result, end up stuck to "safe" web sites, etc.

Which is how we got here in the first place.