i am once again here to say agent discovery & registry hell is one of the least impactful things to focus on that for whatever reason keeps nerd sniping cracked ppl...most "mid" solutions for it are good enough and "better" solutions barely move the needle either you: 1) actually take your 'thought experiment' to its sci-fi logical conclusion wrt "autonomous" + "trustless" agents and make them a first class citizen by enshrining the {fdn model, tool use, etc.} components of an agent directly into the chain* (we do this) rather than frankensteining together some off-chain & on-chain stuff as mentioned in 8004 (we did this: infernet) 2) stay grounded in reality & grok how the largest b2b / b2b2c web2 agent startups (sierra, decagon, ...) that are printing real $$$ work. either they're in the camp of single general purpose agent deployed many times or a statically defined computational graph of how specialized agents communicate. this intellectual masturbatory notion of dynamic graphs of agents discovering each other much less useful than you may think if you're in the camp of web2 cos getting real users. if you're not in this camp, and believe in futuristic settings, then you should do 1). anything in between is worst of both worlds imo. *feel free to @ me but your fave L1 today (eth, solana, monad, ...) doesn't allow for enshrined agents. it's also highly non-trivial for them to do this given it's not in line with their fundamental design. expanding the exec clients' vm to handle fwd passes across oss llms, network calls for tool use, etc. is the easy part. hard part is bypassing replicated exec in consensus for non-deterministic behavior w/out ~degrading safety/liveness, fee mech, and how to allow for scheduled txs to exist in a way without borking the perf / end UX for regular txs. wouldn't be possible for us without the gigabrains @noamnisan @n_durvasula @bahrani_maryam and others coming up with some new machinery. reality is most L1s are an exercise in networking-bound settings...we're in an exec-bound one.

"our market maker fucked us" been meaning to write about this for a while but reality is most founders in the space are retarded at expressing & pricing risk. given the strong experimentation / tinkering culture, naturally makes sense that founders think they have to "innovate" by introducing what they think are clever terms in these MM deals. what follows is classic adverse selection given they're negotiating w/ entities whose explicit job it is to price risk in a structured way, so they'll continually reject -ev terms and not explicitly call out overly fucked terms for the project and claim ignorance later (++ev). usually fucked terms for the project end up being strongly correlated to fucked terms for retail / community / etc. which leads us to where we are today at first glance, soln sounds like to shove transparency down everyone's throats wrt disclosing as many details as possible (i.e. blockworks framework, etc.) and this will obviously help, but even if they were to disclose the exact MM contracts publicly still would be overly misleading, potentially moreso since it sounds like complete transparency but what you really need is a proof of inexistence of any other bilateral/etc. contracts they have made that lets them express risk in a way that's not conducive (more importantly, transparent) to the project... i see a ton of comments in cobie's thread theorizing around skewing greeks (primarily delta) and there's a ton of ways to continue to do that even if you disclose what appears to be a benign MM contract & "their positions". short of you hitting up every credit/structured products/otc/etc. desk and somehow hoping they'll eventually leak any contracts they have w/ the "MM" in question or, instead of getting an "auditor" after the token spirals, sticking as many commlit/tech trans/accountants you can on the MM (or parent conglo) and attempting to get a biospy before the deal goes thru. no one will agree to the latter (and it costs a fuck ton + slow), and if a founder could pull off the former, well, they should probs get a diff job than the one they have. and no, hyperliquid doesn't "solve this" (yes, im sidelined), but actually doesn't given no matter how much you can migrate onchain in a transparent way, showing the inexistence of offchain contracts is what you really want. there have been a few intermediaries as people mention (i.e. coinwatch, glassmarkets, etc.) that attempt to police size/uptime/width in some trust-minimized (TEE/zkTLS/etc.) way of info leakage to MMs but same point still stands re: proving inexistence. also, goes w/out saying, but to all the people that keep asking "why do projects give away so many tokens to MMs" and then come up w/ ways to infinitely police them via trust-minimized "subaccounts" for monitoring-if you gut a MM's capital efficiency, well, that's what tends to happen re: need to juice their incentives given there's an absymal lack of primes in the space and unsecured credit has ~dissipated since the '21 cycle. speaking of the latter, to any one that dealt at scale w/ genesis/blockfi/etc. last cycle, pretty well known the amount of closed term retarded haircut (w/ no rehypo) deals they were agreeing to last cycle—will inevitably come back in true fashion sometime and that will exacerbate policing of greeks that a MM has by an order of magnitude. lastly, if you were around for the mini heyday of arthur's quantos (ethusd, etc.) in the '17 cycle, you'll remember that it was one of the few ways to crudely trade dispersion in the space by proxy of corr but they ended up dying off since hard to price funding, retail choosing vanilla linear/inverse perps over them, etc. but, bringing it back to MM deals & constricting net greeks, given the nature of vol/corr/liq dynamics for quantos, easier to single out idio and not get left holding the bag on getting paid out in shitcoin + better risk mgmt than stables given corr ~beakdown in this sorta market (i.e. get paid in $hype for being 100x short some shitcoin). always was surprised i didnt see more ppl try to rebuild it since it has some tangible pmf...i mean am "glad" @zhusu running the gamut on quantos but ofc 3ac shenanigans last cycle -> doa and fwiw, some of these MM deals ive seen for some projects are overly exotic where it's not even clear any party knows what they're agreeing to but given it's "ok" from a statutory/case law perspective, no attys will flag shit & frankly not well equipped to do so past the basics on commercials. ill give a concrete example—there's at least one coin MM deal in the top 30 coins today that has some odd mix of "best of"/"asian"/"barrier" type characteristics that im sure some of you know, but it's also written in a benign way where it doesn't look like it's a frankenstein of a structure to price out and probs was the result of someone trying to repeatedly bandaid the canonical loan + call option structure w/ anecdotes they heard in the space. and to be clear, it's worked out pretty fucking well for the project but def a few MMs had no clue how to price it and approximated it to what they can price. but, my broader point since the start is, trying to prevent bad things from happening via attempted policing in the absence of invasive biopsies of MMs' external deals won't lead to the intended transparency (or outcomes) ppl want. need to get back to icml but will write more on this later if ppl are interested, but really wish founders would sit down w/ any ~not heavily redteamed reasoning llm, pipe in the contract, and ask it to enumerate diff scenarios for whatever clever structure they think they want as well as net greeks for everyone involved + implications lastly, we'll see all the same clownery from web3 founders this cycle in terms of thinking it's "easy" to clean shells, "easy" to issue converts to everyone so they'll gamma scalp to be the "buyer of last resort", etc. for all the treasury company shenanigans set to ensue