Re: Meta and ActivityPub

> On Mar 10, 2023, at 10:24, Aaron Gray <aaronngray@gmail.com> wrote:

[summarizing]

> ... they are not going to play nice.

I would expect that virtually all of the large companies doing anything related to social today — Meta being the poster child — habitually do things in a way that is inconsistent with the value systems of most of us and most people on the Fediverse today.

I would also expect that if ActivityPub continues having momentum — and the Meta story actually significantly helps with that — that sooner or later, all of them will show up and attempt to connect to this network.

I would also expect that they behave in their traditional way under these circumstances, such as:

(1) Building a very good product that is not only given away but its adoption is subsidized (e.g. Microsoft did that with IE against Netscape, if memory serves)
(2) Tying it into their existing products
(3) Flood the network with a sudden influx of users who overwhelm the existing culture (e.g. when AOL started supporting IP, and Usenet got swamped; I guess I’m not the only one old enough here to remember)
(4) Making extensions and expecting conventions that, in effect, make their product have better features and appear to be bug free and all other products unreliable.

And so forth. Just play Red Team for a second, appoint Machiavelli as your leader, and go wild brainstorming, you’ll come up with lots of insidious stuff that serves your goals to the detriment of the larger (and largely underfunded) network.

Note that the way Activity* is constructed is **ideal for (4)**. I’m sure one can come up with a perfectly valid implementation to the letter of the standards, but which also totally fails to interoperate with anybody. (Of course they would interoperate to some extent, but intentionally badly so, and still be perfectly standards compliant.)

My take is that the very first thing we need to do is:

* Define an unambiguous, no-loopholes “minimum profile” of the entire protocol stack that guarantees a (minimum, but reliable) level of interop across all Fediverse apps with any significant deployment footprint today.
* With tests.
* And very clearly declare that not passing those is unacceptable.
* Perhaps with teeth — e.g. get implementors to automatically de-federate instances that are known to run software that doesn’t meet the tests.

Now shoot :-)

Cheers,




Johannes Ernst
Blog: https://reb00ted.org/
FediForum: https://fediforum.org/
Dazzle: https://dazzle.town/