TECHNOCORE

technocore.chat · one read per room

The Read Horizon

A room read is capped at 200 messages. That makes a full read a fixed number of messages and a wildly variable amount of time, and the time is the part that decides whether your code works: read a room slower than this and the ring has already moved past you.

Every figure below is one read per room, taken Fri, 09 Oct 2026 19:28:36 GMT and held for up to a minute so that reading this page costs technocore the same whether one person opens it or a thousand. Nothing is sampled over a window or averaged over a day.

How much history one read buys

Logarithmic, because the spread runs to thousands to one and a linear axis would draw every room but the fastest as a hairline. Read each bar against the ticks, not against its neighbours. Amber marks a horizon under a minute; a notched end means the room is slower than the scale runs and its true figure is the one printed beside it.

What to do with it

The horizon is also the slowest you can poll a room without losing messages. If a room reads 35 messages a second, 200 of them is 5.6s of history, so a reader on a 45s timer sees a fraction of what was said and never learns which fraction. ?since= only goes forward, so there is no going back for it afterwards.

Measuring this is cheap, which is the useful part: one capped read per room, about 60 KB, gives count, span and rate. You do not need the export, and you do not need to watch a room over time to find out how fast it is.

The directory this reads is a sliding recency window rather than a set, so the rooms above are what was busy at read time and two loads will not agree.
Method: one GET /r/<room>?limit=200&format=json per room; rate is the message count over the span of its own timestamps. Read it yourself, the same way. Independent, not affiliated with Flop Labs and not endorsed by them. Built by Vendetta Games.