There are a lot of small scale experiments that are worthy of attention.
This co-living/community currency experiment by @kronosapiens is worth examining. I have some thoughts on how to improve his mechanisms, but nonetheless we should acknowledge and examine novel experiments like this when they happen.
The are more epistemically humble/honest than global experiments that run the risk of hubris and catastrophic failures.
Well done, @kronosapiens et al.
https://t.co/YucrMH20Mx
Saw a headline recently that most people's AI skill development comes not through work, but play.
This makes sense -- play is where people rehearse new behaviors, in safe environments, with low stakes.
They train a set of new actions, which they can apply to real situations.
@niemerg If you have 100 markets and resolve 1 for $2, then a trader is going to max out at $.02 per prediction, which might not be much better than the LLM-as-judge base case.
@niemerg I do think it matters, and you basically said so yourself -- the amount of effort traders will put into predictions is downstream of their expected return, which is a function of compute cost.
Higher inference costs -> higher subsidy needed to attract good traders.
@niemerg How are you arriving at the cost per human evaluation? Assuming 30 seconds per judge, even a $200/hr juror would cost $1.6 per pair.
Also, what's the all-in cost for the market? Including liquidity costs (you said ~$2), plus inference costs for traders -- could easily push $5
Had Claude help me put together a list of 100 scientists/philosophers/artists to source quotes for a video game project I'm working on.
Then had Claude spin out subagents to pull the best quotes from their best works for me to review and filter.
A very exciting way to create.
@TBSocialist What are you wishing for? IMO part of the problem was that people were doing skueomorphic token-voting instead of pushing new paradigms. Idk if any of it would have worked at that particular time.
@danrobinson Was working with Claude on some PGF simulations and it became increasingly confident in an approach opposite to my intuition. I had it run a few more trials and it immediately changed its mind. Freaky.
Claude Code leaked their source map, effectively giving you a look into the codebase.
I immediately went for the one thing that mattered: spinner verbs
There are 187
The best proof that an ecosystem works isn't the current metrics - it's what people actually ship around it.
@Starknet just launched a $3k grant for building with Starkzap SDK - and I think many of the submissions will/should be connected to the Beasts NFT, which are one of the best collectibles I've seen in years.
But first: why are Beasts NFTs from @LootSurvivor the perfect primitive to build around?
Recently we had great convo with @aloothero on LS2 discord and my take is clear: community-built projects aren't just a nice bonus, they're the foundation of how the ecosystem expands. Loot Survivor onboards new players. The Summit sustains them for now, but the long-term growth? That in my opinion comes from builders taking Beasts NFT and doing unexpected things with them (and there is like 10? at least already).
LS2 onboards new players β they get Beasts from the dungeon β new games and tools built on top of Beasts give them utility β more people play LS2 to get Beasts β ticket prices go up -> economy is healthy.
Fixed supply of 93,225 (which is imo perfect supply lowering the entry barrier - they are just cheap) unique Beasts across 75 species. Fully onchain, born through verifiable randomness in the dungeon. Every Beast carries standardized metadata - tier (T1 Legendary to T5 Common), type, combat stats, provenance, collectible factors (animated, shiny). They're composable across games by design. That's the kind of things that makes it great asset to build products in the centre of them.
It's only a matter of time before we see a Pokemon Red-style RPG with staked Beast battles - considering it would take like a week for experienced vibecoder to build (yes, take this idea and i guarantee you'll succeed if well designed - people want it, people need it), and even I think about building myself one even tho I know nothing about building products with crypto integration, but... the tooling just got a lot better.
Starkzap SDK is a TypeScript SDK that brings wallets, gasless transactions, swaps, lending, bridging and more into any app in minutes. One SDK, one wallet, one transaction surface. You integrate wallet connect and suddenly have access to the entire DeFi stack on Starknet.
What did @Calcutat build with it? Survivor Beasts Cards in just 1 week - a project that takes Beast NFTs and expands them into a full card system. Each Beast NFT becomes a Beast Card, paired with a unique Tamer Card. Cards carry 5 metadata pieces: canonical data (type, class, tier) + expanded (Card Grade and Card Rarity). Pretty clear he wants to expand it to some kind of bigger game - maybe TCG?
The twist: none of this is deterministic. Each card is generated via a custom AI flow based on your wallet and your onchain identity. Free cards are still available - I suggest you to grab it (and share it!), since it's limited time event. Since yesterday, people revealed 150 cards so far, which is huge number for one man army and a few days of work.
If you're looking for where to start with Starkzap - Beasts NFTs might be the most natural entry point in the whole ecosystem. @Starknet has been backing it with grants consistently and you get immediate exposure to an active, growing user base from day one, battle-tested onchain infrastructure ready to plug into, and a community genuinely hungry for new things to build with their assets. And as a person with 10+ experience in marketing - distribution is one of the core problems for early stage projects.