Imo Xi's idea about 100K Americans studying in China is excellent. I'm real tired of the whining about "they steal our secrets at Our Amazing Schools, then go back…". You just can't learn Mandarin nor pass the gaokao lol. This would both help American skills and reduce hubris.

IMO AI enables the creation of new marketplace in healthcare in a few different ways 1) It makes inventory more discoverable - This can be knowing what's in stock at the pharmacy, what appointments are available, if a doctor still takes patients with X insurance, what kinds of patients are in the panel, etc. You can now verify this at scale with things like calls, structuring data in the EHR, etc. 2) Agents can match suppliers and buyers more easily - Agents can take requirements that buyers have match them to sellers even if it's a fuzzier match. Before you're need to create really specific tags to make that match happen, but now you can do it on pretty much anything. This becomes even more complex if the transaction requires 3+ entities. 3) You can bootstrap new networks quickly - If a company isn't in your network, you can automate a fax or call to transmit the same info. Using existing rails that anyone can participate in basically means anyone can be in your network. Examples of new marketplaces I'm personally invested in: Photon - better marketplace for providers, pharmacies, and patients by making pharmacy shoppable Turquoise Health - better marketplace between payers, providers, and employers by turning negotiated rate contracts into something more modifiable + able to be transacted on top of Double Blind Bio - better marketplace between sites and sponsors so they can find each other based on their needs I think there are many more to be built here

imo the millennium problem list in biology should be a list of very specific datasets like ‘a catalog of epitope-resolved nm protein binders across the human proteoform in X environment’. hard because it requires knowing the full range of human ptms/isoforms and having some way to do scalable generation/testing of epitope-specific binders (computationally or otherwise). both very difficult but achievable also verifiable!

IMO the essential difficulty of malleable software now boils down to “fork maintenance”. Let’s unpack that a bit. If you want to make your own custom tools or fork existing OSS, the cost of that is headed to zero. But… often we don’t want to start from scratch! We want to remix tools others have created. We want someone else to help us maintain stuff. We want to combine our ideas with others’. Now the moment you have multiple parties both updating the software… you hit gnarly problems! In a nutshell: happens if I modded something and it no longer meshes well with the latest updates from you? This is “fork maintenance”. (Traditionally this would be a centralized model where a “dev” is maintaining the software and you, the “user” might adjust settings or patch things. But even in a more decentralized setting we want to jam with our peers on our tools.) A classic approach here is to expose a stable plugin SDK. Works great because the dev can keep the surface maintained and plugins hopefully compose reasonably…. But only works as long as the plugin surface has what you need. Sometimes you end up hitting walls. If it’s OSS you can also literally fork. Maintaining a fork and merging edits from others is a huge pain. But maybe with enough smart cheap LLMs this is more viable now. You also need to consider data, not just code. Schema migrations are already a pain in the ass with normal software; when every user is doing customizations things get extra gnarly. I don’t think that there’s a perfect solution to this yet. Ideally I’d want something much deeper and broader than a typical plugin API - eg a language or framework designed for layered additive modification (think Harel’s behavioral programming as one reference.) I find it quite poetic that collaboration and working with others is where malleability gets really good but also where it gets hard. Same as many other things in life! For more on this, we go into quite some depth on this topic here: