with oil prices plummeting amid the middle east conflict, looks like markets are becoming an early artifact of new information asymmetries. exciting times to watch capital and intelligence converge www.ft.com/content/611f...
At Heroku, we basically destroyed a generation of folks personal projects when we deprecated various free plans. (My own software died too.) Now Glitch is making the hard decision to do the same: blog.glitch.com/post/changes...
The lesson is simple: don't build in the cloud if you want it to last.
come say hi! great opportunity to meet other creative people working at the frontier of LLM code gen to retake the role of technology as an aid for daily life
what people aren’t considering that this is probably less hallucinatory than the information satya normally gets
in the meantime, before infighting about who's more decentralised, articulate the specific benefits you're after and go from there. this is far more productive than arguing over presumed and ever-shifting goalposts. 12/12
this shift would help us evaluate technologies based on how they actually empower users rather than whether they match our definition de jour of "decentralised." 11/12
reorientation doesn't mean abandoning the underlying values, but rather spreading them across a more information-rich gradient that helps us be precise about what we're trying to achieve and how architectural choices support those aims. 10/12
decentralisation as our north star never made sense, but now it's becoming obviously counterproductive. while the energy pouring into networked tech is fantastic, we need to align ourselves along a shared, delineated taxonomy, enabling more productive conversations and clearer intuition. 9/12
the centralisation/decentralisation binary is a false dichotomy, obscuring that most successful systems operate on a spectrum, and indeed one often necessitates the other. 8/12
activitypub, for instance, provides some resilience and user choice while compromising on UX for the uninitiated. but a federated system that doesn't prioritise interoperability might be worse than a centralised one with good data portability. 7/12
sometimes decentralised architectures help us towards these goals. often, they introduce new problems: poor UX, higher technical barriers to entry, inefficient resource allocation, etc. 6/12
what we actually want are systems that prioritise:
- free flow of information
- user agency and control
- interoperability/composability
- censorship resistance
- resilience against points of failure
- and plenty more
5/12
this makes decentralisation an end in itself rather than a means. we forget that decentralisation (in all its definitions) is one possible—complimentary—approach to achieving more fundamental values. 4/12
"decentralisation" has been stretched to encompass everything from bitcoin to the fediverse to P2P, all with wildly different architectures. the word is now a stand-in for all that's "good" or "desirable" in future-facing technologies, obfuscating the actual value of the underlying tech/vision. 3/12
this problem is particularly evident between proponents of atproto and activitypub. the network topologies are vastly different, with explicitly separate aims. so why do we continue to argue over custody of an over-defined term in an under-defined field? 2/12
this dialogue perfectly captures the problem with "decentralisation" as our guiding principle. we've reached a point where the word has become nearly meaningless through overuse and inconsistent definition. the goalposts constantly move, making productive conversation impossible. 1/12