A civically robust information-sharing and discussion system.

The use of overlapping Webs of Trust with well defined meanings to appoint the curators of content tags in a way that is easy to grow, easy to collectively moderate, and when it needs to be, subjective. As far as I can tell, this will make content tags much more practical than they previously have been, and it'll enable a whole lot of really important conversations and community processes that weren't possible before.

A type of multiplayer game in which peace is possible, but difficult. Games that teach negotiation. The core mechanism tends to be, players are trying to optimize their individual scores rather than strictly trying to outscore other players, this is a subtle distinction, but it has great consequences:

It models an agent's utility functions in a way that's much more true to life than purely positional (relative) scoring is.

It means that, although there is conflict, one person's gain is not always another person's loss, beneficial trade is possible, peace is possible, honest negotiation is possible. For me this makes peacewagers much more interesting and relevant to the real world than the zero-sum games we're used to.

A vivid, detailed, plausible rendition of a future-history in which our world meets with the worst contingencies in the development of AI, and survives. The first-winning entry in the Future of Life Institute's 2022 positive worldbuilding contest.

Told through short stories, a timeline of events, and a Q&A.

It discusses the scenario where capabilities arrive abruptly but alignment is hard, in which the world needs to manage unprecedented levels of global coordination. I propose that humanity will not survive this outcome unless it's prepared to very quickly produce clear demonstrations of the presence of danger.

So I outline a technical programme for relatively safely demonstrating the danger, and an international response to it. I have doubts about these solutions, there are always doubts, we'll just have to sit down and do the work if we want to see whether they're resolvable.

There's also a bit of hopepunk at the end, depictions of the kind of world humanity stands to win.

See also, the appendix, a non-fictional exploration of some of the core ideas, with much more thorough discussion of the technical program.

A short story/thought experiment/speculative scifi about why we need a concept of anthropic measure that's related to but conceptually distinct from the agency behaviors of consciousness. It wasn't totally clear to me that there existed a hard problem of consciousness until I found this thing. I think it clarifies a lot.

the story

tweets

A Glance Font. A font reduced to the simplest and most pronounced shapes. I made bokik to test the hypothesis that a fluent reader of a glance font would be able to read faster or with fewer errors (or assumptions) than can be had with traditional typefaces. As far as I'm aware, no one has ever become a fluent reader. Please tell me if this changes.

It looks like this. I expect that becoming fluent in bokik wont be easy, people don't read letter by letter, they read word by word, so it doesn't just require you to learn 26 things, it may end up requiring you to learn around 50,000.

Utilitarian laws for a fluid type of city where land is allocated in whatever suggested arrangement minimizes unnecessary moves while maximizing propinquity (good adjacency) according ticketholders' (residents) desire expressions and the city's aggregation law, obviating rent, restructuring the city to optimize the needs of its communities (occasionally forming new communities), enabling much better standards of living.

Anyone with insight into numerical modeling, voting theory, policy communication, construction or urban economics please get in touch and help me to either refute or implement it.

It's difficult to build a great city when it is primarily owned by uncoordinated, unaccountable individuals with relatively little interest in the health of the whole. Proq offers an alternative.