snarfed

The other key thing you lose is your identity. In the fediverse, your identity is your Webfinger handle, ie @[email protected]. Your server is literally part of it. Sure, you can migrate to @[email protected], and keep the username part, but your identity still changes.

Truly portable identity via DIDs, ie you can keep your underlying identity even if you migrate servers, is one of the key reasons the Bluesky team made their own protocol. https://atproto.com/guides/faq#why-not-use-activitypub

Love it! I keep a similar set of quotes. Lots of fun. https://snarfed.org/software-engineering-quotes

This has demo scene written all over it, in the best possible way.

This is https://tidelift.com/ ! Others too, I think.

Not recent but legendary: "Latency Lags Bandwidth" David Patterson http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.115...

Recent:

"How to Hack the Simulation?" Roman Yampolskiy https://www.researchgate.net/publication/364811408_How_to_Ha...

"On the Computational Practicality of Private Information Retrieval" Radu Sion, Bogdan Carbunar https://zxr.io/research/sion2007pir.pdf

(via "Explained from scratch: private information retrieval using homomorphic encryption," https://blintzbase.com/posts/pir-and-fhe-from-scratch/ )

OP here. Definitely! Both of those are important too, SSB especially, it's been around for so long. Plus Rabble has been vocal in fediverse and Bluesky development and now actively building Nostr stuff.

(And I'm actually all for multiple protocols, especially right now when we're still fairly early in the space's mainstream growth. Much better to experiment with lots of different ideas now while we still have a bit of breathing room!)

"silos as plumbing"

Definitely the same era! Looks like COPE was 2009, POSSE concept was 2010 https://indieweb.org/timeline#2010 , term was 2012 https://indieweb.org/timeline#2012 .

You don't have to automate it. "Manual until it hurts."

OP here. Sure, they finally anounced the new pricing, months late, but all of the evidence in the post that it's effectively unmaintained still stands. The hostile new pricing just adds to the lack of support in convincing me not to waste any more of my time developing for it.

snewman's comments have addressed most of this. (hey steve!) i'll just add that HRD has more predictable read and write latency, which most developers like a lot, but its writes are also slower on average, since they incur at least one Paxos round across datacenters.