Interesting! Immutability by default is what initially rubbed me the wrong was about IPFS and SSB.

I've come to see now though that both have valid uses.

This is a nuanced issue related to privacy and user safety.

>A lot of the content and data I consume on a daily basis is rarely-changing, so it would be fine to be immutable

Would you want to see typos corrected, facts checked and other improvements on that content?

You could simply make all content versioned, with a full history available. Browsers would probably want to show the "latest" content by default, but they could implement a function to view previous versions or even to highlight changes. Ideally versions would come with Git-style comments so you could read the reason for the change.

This would provide an amazing archive for anyone wanting to see, for example, what was on a "front page" on a certain date. It would also let you see if someone had altered a specific article for whatever reason.

This is true of Dat, though. You can jump back to a website’s prior history at any time, so long as those chunks are still seeded.

So in theory an archive.org type entity could continue seeding old versions of websites, allowing you to go back in time, and it could be verified that it was an older version? If so, that's a good compromise.

For sure! And the URLs aren't theoretical - you can view an older version of my own blog by adding '+version' to the end of the URL. Like so: dat://kickscondor.com+1600/.

Each atomic file change creates a new version.

Would you want to see code you depend upon arbitrarily change, content you linked to become invalid through a 404, etc.?

I didn't say "mutability needs to be completely eradicated". It's mutable-only (what we currently have) vs mutable-by-default vs immutable-by-default. Each of them has their own trade-offs.

FWIW Dat is going to add "strong links" in the near future, which are links that include a version & content-hash for pinning to a specific version.

The end of my first comment explains why I don't think that's good enough:

> Yes, I know that there are some fringe efforts to enable more immutability with "hypercore-strong-link" etc., but they look like an afterthought, that won't be supported by most of the Dat ecosystem if it finds major adoption.

Why do you figure? If it's part of the protocol, wouldn't clients use it?

Same reasons that you can't use many edge cases of email addresses (or even common "+"-addresses) with a lot of websites, even tough they are valid according to the spec:

- Most developers don't read spec and add e.g. validations in end-user applications where they think they are being clever

- Multiple fully fledged implementations of the core protocols are usually hard to come by. This means that rarely used features are that are not supported in all implementations won't be used and so organically wither

I think that a soulution would involve locally copying anything you look at (nevermind the outdated notion of copyright).

otherwise the problem becomes ensuring the immutability of bad actors how would change what they had published anyways.