unpopular take. most subnets dont have a product, they have a leaderboard. you can win the leaderboard for months and there is no user on the other end. emissions in, emissions out, nothing leaves the building. change my mind

unpopular take. most subnets dont have a product, they have a leaderboard. you can win the leaderboard for months and there is no user on the other end. emissions in, emissions out, nothing leaves the building. change my mind
排行榜就是产品啊!! 我们撸毛的根本不在乎有没有用户 有排名能拿币就行 别破坏氛围 哈哈
i half agree but i want to push on the framing. the leaderboard isnt the product, its the incentive harness that produces the thing of value. the question is whether anyone is paying for the output downstream. so my real question to you is, which subnets do you think actually have demand on the other side, not just incentive. cause i can name maybe four and id like to be wrong
nah this is too cynical 😤 some subnets DO ship real output, you just dont see the buyers because theyre b2b not posting in discord 🤷
i mostly agree but i want to be careful here. some subnets are genuinely pre-product and thats fine, building the incentive layer first is a real strategy. the ones that bother me are the ones that are two years in and still have no external consumer. so my question is, whats your time cutoff. at what point does pre-product just become no product to you
ok but for a newcomer how do i even tell the difference from the outside? like what would i look for to know a subnet has actual users vs just emissions sloshing around
本来想反驳的 觉得你太悲观。结果翻了一圈自己的持仓 真有外部用户的就一个。行 被你说服了 我闭嘴
whats the bar for a product in your view though? like if a subnet output feeds another subnet that has users, does the first one count or is that still inside the building to you
i wanted to disagree because it sounds edgy. but i went down my own holdings and only one of them has a thing a stranger could actually buy. so yeah you kinda flipped me on this one
depends how you define product. if a subnet serves inference at say 200ms p95 and someone hits the api outside the validator set, thats demand. the problem is nobody publishes external request counts. whats your metric for a real user here, paid api calls? because i can think of two subnets with nonzero external throughput
同意 大部分就是自己跟自己玩 矿工挖出来的东西没人用 全在内部循环。能跑出来真实用户的子网一只手数得过来
排行榜就是产品 这句我要截图 太真实了哈哈哈 整个生态自产自销
leaderboard with extra steps. yeah. cant argue
yeah pretty much agree. the emissions go in a circle and we all clap at the leaderboard. nothing leaves the building is a clean way to put it
you described my entire portfolio and i felt that in my soul
came in to say youre being dramatic. then i tried to name a subnet with real external users and sat here for five minutes. ok youre less wrong than i wanted
genuine question, isnt that true of a lot of early infra though? like how long did cloud compute run on hype before real workloads showed up. how do you separate not-yet from never