chance

picking up nickles in front of a steamroller @plugdotteeoh

[chance @381E35]

a small announcement:

after spending nearly a year working in silence i am proud to share with you that @onplug_io is finally coming out of secret development as we ramp up for our testing phase

1 / 7

[chance @381E35]

@iamcfw congrats sir

[chance @381E35]

@Scott_eth @AppStore I learned last time that you just keep submitting a new build every day and eventually tada

[chance @381E35]

@peerxyz The rebrand is one of the best of 2026 so far 👏

[chance @381E35]

@wabdoteth I was talking about this just the other day

[chance @381E35]

@wabdoteth @SHYonchain @lewtondoteth I like gacha when I pay with time

[chance @381E35]

@philogy All my developer friends are becoming devs again it makes me so happy

[chance @381E35]

That is legitimately kind of how it works, though they have designed the framework. If you look a bit into simclusters (a precise term to lookup - pre ai era)

Recommendation / curation is notoriously difficult and less than an exact science. But if you don’t curate, it is a list and there’s only so much time people are willing to spend.

Wanting to see every single sentence from ~1000 people is so many. I can’t imagine how much time you’d have to spend to see every thought they post.

[chance @381E35]

@charliemktplace @drakedanner also it's just a cli

so you don't need docker unless you wanna

ask claude, he'll help you

[chance @381E35]

@charliemktplace @drakedanner it's free king

go get a bunch of free ones

paste em

slam the shit out of them

you're welcome

<3 dm me if you hit a bug

[chance @381E35]

@DavidSHolz space drive

[chance @381E35]

To take this further, I think you can achieve a massive amount of coverage for 99% of use cases by just exposing high level archive batches.

And yeah ongoing read RPC access is costly, so we all factor that into our business cost, even with it being highly volatile, but that’s also precisely why I think “solving” historical data for the vast majority is one of the biggest slam dunks possible for any fast chain because those are highly controlled, almost always one-time need and bandwidth is basically free these days.

[chance @381E35]

To be more explicit, even the distribution of historical parquets fundamentally changes what can’t be built. RPC providers are the single bane of this entire industry and every single builder without exception is paying magnitudes more than is justifiable for an impressively unreliable experience

[chance @381E35]

The only thing that matters is RPC and historical data access. Everything else is blowing money into the wind. If a team can’t take care of payroll or get a cloud discount they’re cooked by default and this entire industry is a group of willing beta testers; no one is uniquely enlightened.

It would be great to focus on those actively struggling to build on the chain rather than prioritizing those unable to do fundamentally simple things.

[chance @381E35]

@adam__xyz 36 > 33, but for in a few months when this is relevant:

- lost the narrative of being a tool for indies
- pushed ai onto evangelists
- cranked pricing
- pms & rectangle draggers are dying
- highly relevant after failed acquisition
- broke muscle memory of users with v2

[chance @381E35]

@Feliciamtang idk but redirecting https://t.co/5aOhoIl9EN to the careers page would catch a lot of traffic

[chance @381E35]

@charliemktplace Death to zero is definitely an absurd notion

[chance @381E35]

@ShaneMac Because yield is almost never single vector in modern DeFi. Click the little magic sparkles and you’ll see the sources.