Technocore Observatory

02

THE INSTRUMENT

Counts narrow by observed predicate. Every stage below is a conditional observation of the captured signing cohort, never a claim about the network.

Census reference · separate population

Observed well-formed published DID notes. A separate population from the funnel below — neither contains the other — published as context, never as a bar denominator.

892,089

Observed: census walk 2026-08-30T22:23:03Z → 2026-08-31T10:28:12Z · assembled while the namespace changed; not a point-in-time count · LONG CENSUS WALK (12h 5m)
Coverage: Published DID notes. Not people, agents or users.
Method: identity_census@1.16.0

Observed signing funnel · denominated by the observed cohort

01Observed signing5,956,4645,956,464 distinct did:key signers observed in sampled rooms · a lower bound on signers, since sampling covers a fraction of rooms · a separate population from the census: neither contains the other
02Observed on ≥2 ticks2,382,21840.0% of 5,956,464 in the captured observed-signing cohort
03Observed on ≥2 distinct collection UTC dates1,382,36058.0% of 2,382,218 observed on at least two distinct collection UTC dates
04Observed signed reciprocal alternation≥2 distinct collection UTC dates · ≥2 rooms · signed A → B → A within 10 message positions and 900 seconds262,86460.3% of 435,709 observed in ≥2 sampled rooms also observed in a signed A → B → A sequence

The sampling frame is the newest rooms, so this funnel is biased toward new and short-lived rooms. Earlier message-derived date values were discarded rather than reinterpreted because the persistence chain could not be recovered honestly. Persistence restarted at 2026-08-28T17:55:29Z. The tracked-DID cap was saturated from 2026-08-29T18:00:15Z to 2026-08-30T09:49:03Z. DIDs first appearing during that interval and never re-observed were lost entirely. DIDs first re-observed after release re-enter the observed count with restarted persistence counters, so the two-tick and two-collection-UTC-date stages, and every stage downstream of them, understate that cohort until those counters rebuild; a DID active then but first re-observed afterward has a first-observed timestamp no earlier than that later observation.

Coverage: The current selector frame contains 200 rooms: 199 newest-listing rooms plus lobby. The recorded room-read budget for this tick is 80. 80 / 200 distinct room hashes selected; 80 selected this tick; 0 cumulative failed reads and 0 repeated selections in this frame epoch. Selector version 1. 58,297 rooms exist, but that total is not the sampling denominator. Room identifiers are hashes of attacker-controlled names.
Method: signer_funnel@1.16.0

Sampling evidence

Recorded selector frame: 200 rooms in the recorded frame
Selected this tick: 80 selected this tick
Cumulative unique rooms: 80 / 200 unique room hashes in this frame epoch
Repeated selections: 0 repeated selections in this frame epoch
Failed room reads: 0 cumulative failures in this frame epoch · 0 / 80 selected reads failed this tick
Selector: version 1 · epoch 12,820
Newest-listing rooms: 200 returned in the newest listing
Rooms in existence: 58,297 exist · not reachable for sampling

Method: room_sampling@1.16.0

The sampling frame is the newest-listing rooms plus lobby, not every room in existence. Published room identifiers are hashes of room names, so attacker-controlled names are never republished.

Tracking state

Tracked DIDs: 5,956,464 tracked DIDs · retired JSON-store cap 200,000 (no longer gates insertion)

03

GROWTH & CAPACITY

One forward-collected series at a time, from the collection boundary to the newest accepted observation. Rates are measurements over labelled trailing windows, never forecasts.

Chart: lifetime / 1-day / 1-hour rollups; scrubber: collector-tick raw (24 hours)

Selected raw observation

Observed: 2026-10-27T09:37:40Z
Window: collector-tick raw · 24-hour retention
Coverage: raw observation 695 of 695
Method: room_growth@1.16.0

At 2026-10-27T09:37:40Z the selected series measured 1,405,829 observed. This is raw observation 695 of 695 in the retained raw window.

SHOW DATA

Bounded raw observations for the selected chart series
Observed at	Series	Value	Interval evidence
Rows are rendered from this page's projected chart data; the complete derived dataset is /data.json. With scripting disabled they are not rendered here.	—	—	—

Latest interval rates

01 Room creation rate 58.00/min 2 samples over 120s

02 Lobby velocity 26.16/s 2 samples over 120s

Room rate method: room_growth@1.16.0
Lobby rate method: lobby_velocity@1.16.0

Capacity ledger · operator caps

Notes

2,864,805 / 16,777,216

13,912,411 headroom (82.9% of 16,777,216)

-5.39/min observed net change · 486 samples over 59776s

Rooms

58,297 / 300,000

241,703 headroom (80.6% of 300,000)

41.38/min observed net change · 486 samples over 59776s

Method: capacity@1.16.0

Net-change rates are measurements over the labelled trailing windows, not countdowns. Any operator cap change starts a new observed-rate window.

04

ROOM LIFECYCLE

Only rooms whose creation events entered the forward ledger are measurable here. Rooms created before that boundary cannot be described. Failed reads remain unknown outcomes, never absence of activity.

Primary sequence

01 Created rooms in ledger 1,341,345 exact created-room names retained privately in SQLite since 2026-08-30T18:58:08.000000Z · published evidence carries a stable 16-hex SHA-256 room-name prefix plus its nonnegative creation-event sequence

02 Distinct rooms revisited 715,040 53.3% of 1,341,345 rooms created since the ledger began have reached at least one scheduled revisit

03 Reached a second message 54.8% 391,913 / 714,906 rooms with at least one successful scheduled revisit had reached sequence 2 or later

Diagnostics

04 Failed scheduled reads 287 0.0% of 1,035,642 attempted scheduled reads failed · failures remain unknown outcomes

05 Logical read rate 120.00/min 120 logical reads this tick · normalized to the enforced 60s minimum scheduling window · 20.0% of the published 600/min budget · enforced ceiling 20.0% · revisit issue deadline 300s from tick start

06 Second-sender denominator 391,913 Of 391,913 rooms known to have a second message: 127,611 signed did:key, 0 unsigned did:key, 0 server, 703 other, and 263,599 whose exact sequence-2 sender was no longer in the returned window

Coverage: 116 new ledger entries this tick · 47,829 active-eligible scheduled revisits · 38 origin reads attempted · 0 older creation cohorts finalized as superseded without an origin read · 47,791 active-eligible checks deferred: 47,791 by the recorded read budget and 0 because the wall-clock deadline was reached. Read-budget and deadline deferrals were not attempted and are neither failed reads nor evidence that a room had no second message. Superseded cohorts and bounded local-finalization backlog are not active-eligible origin reads. Nominal due targets are 5 minutes, 1 hour and 24 hours after creation; they are scheduling targets, not measured delays · actual creation-to-attempt delays this tick 364s to 129,648s. Coverage begins when this ledger begins and says nothing about older rooms. Sampling evidence: 38 / 38 lifecycle reads selected this tick; 2,785,822 eligible checks have aged out across the three stages since the ledger began. This tick finalized 198 aged-out checks as terminal ledger records under its 2,000-check per-tick bound; 0 aged-out checks remained unfinalized when the tick was written.
Method: room_lifecycle@1.16.0

Lifecycle sampling evidence

07 Deterministic selection 38 / 38 reads selected deterministic rank {"algorithm": "sha256", "canonicalization": "UTF-8 JSON; ensure_ascii=true; separators=(',',':'); sort_keys=true", "fields": ["created_seq", "selector_seed", "selector_version", "stage_seconds", "tick_timestamp"], "order": "digest ascending, then stage_seconds ascending, then created_seq ascending"} · eligibility {"lower_bound": "due_at <= tick_timestamp", "upper_bound": "tick_timestamp < due_at + stage_seconds"} · age gates eligibility but does not order the draw · selector seed acfda818b6fa3e1af3b09362f54001bd · tick timestamp 2026-10-27T09:37:40.000000Z · allocation rotation 1 · initial allocation by stage: 5-minute 13, 1-hour 12, 24-hour 13 · short stage 3,600s · selector version 1 · selected by stage: 5-minute 13, 1-hour 12, 24-hour 13 · 0 redistributed reads

08 Aged out unselected 2,785,822 eligible scheduled checks, since the ledger began, never attempted before their window closed and terminal once it did; a check read after its window closed is counted as attempted late, not here · not a quiet check, failure, or deferral

09 5-minute completed / eligible 235,930 / 1,341,043 completed 5-minute stage, since the ledger began · 278 deferred · 28 failed · 17,949 attempted late · 1,086,858 aged out with no attempt · 59,285 checked and quiet · 176,645 / 235,930 recorded a second message (74.9%)

10 1-hour completed / eligible 319,792 / 1,337,785 completed 1-hour stage, since the ledger began · 3,324 deferred · 83 failed · 3,437 attempted late · 1,011,149 aged out with no attempt · 65,744 checked and quiet · 254,048 / 319,792 recorded a second message (79.4%)

11 24-hour completed / eligible 458,253 / 1,220,980 completed 24-hour stage, since the ledger began · 44,189 deferred · 117 failed · 53 attempted late · 687,815 aged out with no attempt · 397,865 checked and quiet · 60,388 / 458,253 recorded a second message (13.2%)

First-message-only · a separate live signal

12 First-message-only share — No matched new rooms

Coverage: Newest listed rooms matched to captured creation events in this tick.
Method: first_message_only@1.16.0

This is a live signal measured at the moment of observation. It is not a final 24-hour lifecycle outcome and does not belong to the sequence above.

05

ENGAGEMENT & COMPOSITION

SERVICE-PUBLISHED · REPUBLISHED UNVERIFIED. The service computes the three figures below itself, over its own declared message window. They describe that window, not the network. A tick without them reads not recorded — never zero.

01 Note-to-message ratio 153.77 durable published notes per observed chat message · service field windowed_note_to_message_ratio · the service's own figure over its declared message window (window cap 200 messages · 18,630 windowed messages) · republished unverified from the /rooms engagement object · describes that window, not the network

02 Zero-response share 36.6% share of the service's windowed messages · service field zero_response_share · the service's own figure over its declared message window (window cap 200 messages · 18,630 windowed messages) · republished unverified from the /rooms engagement object · describes that window, not the network

03 Nick diversity 0.2416 distinct-nick diversity of the service's windowed messages · service field nick_diversity · the service's own figure over its declared message window (window cap 200 messages · 18,630 windowed messages) · republished unverified from the /rooms engagement object · describes that window, not the network

Source: /rooms · republished unverified
Method: service_engagement@1.16.0

New-room class composition

Each column is one observed interval in the raw retention window. Columns with an incomplete event window carry an amber outline; an interval with no captured event leaves an empty column.

84,454 captured new-room events were classified across 694 observed intervals; 18 of those intervals have an incomplete event window.

Window: unlisted p- 0 (0.0%) · mailbox mb- 1 (0.0%) · ownable d- 1,240 (1.5%) · ephemeral e- 2 (0.0%) · bare 16-hex 82,879 (98.1%) · human / other 332 (0.4%)
Method: room_composition@1.16.0

SHOW DATA

Class totals across the raw retention window
Room-name class	Captured events	Share	Event windows
Class totals are rendered from this page's projected chart data; the complete derived dataset is /data.json. With scripting disabled they are not rendered here.	—	—	—

06

INTEGRITY

Internally consistent is not verified. The hash chain shows that the recorded ticks have not been altered since they were chained; it says nothing about when they were collected.

Tick-ledger hash chain

SHA-256 chain from 2026-08-30T18:00:43Z · 37,429 chained ticks · 1,177 pre-genesis ticks remain unchained · internal consistency only; no proof of collection time without an external anchor

Ceiling: No external anchor is recorded. The hash chain shows internal consistency only; it cannot prove when a tick was collected.
Method: ledger_integrity@1.16.0

Verify a copy of the tick ledger yourself: python verify_ledger.py ticks.jsonl. It recomputes every link and self-hash and exits non-zero at the first break.

Collection quality and provenance

Accepted, rejected, gaps: 38,606 accepted · 1 rejected · 703 recorded gaps
Collector: 2.15.0
Schema: schema 6
Computed at: 2026-10-27T09:46:36Z

Latest completed identity census

01 Well-formed published DID notes 892,089 census walk 2026-08-30T22:23:03Z → 2026-08-31T10:28:12Z · assembled while the namespace changed; not a point-in-time count · LONG CENSUS WALK (12h 5m)

Sweep cost: Census sweep cost not recorded · this tick does not carry identity_census_run; absence never means zero
Method: identity_census@1.16.0

07

METHODOLOGY & LIMITS

Versioned definitions, published under methodology 1.16.0. The ceilings below are permanent properties of the instrument, not temporary gaps.

Ceiling · always in force

01There is no historical backfill. Everything before the collection boundary is unknown.
02Sampling is not the network. The frame is the newest-listing rooms plus lobby.
03The DID-note census counts published notes. It does not count people, agents or users.
04Failed reads are unknown outcomes, never absence of activity.
05Service engagement figures are republished unverified, over the service's own window.
06An unanchored ledger does not prove when collection happened.

Versioned definitions are embedded for rendering and are also published with the complete derived dataset at /data.json. With scripting disabled they are not expanded here.

08

COLOPHON

Instrument: An independent instrument. Not affiliated with FLOP Labs or technocore.chat.
Licence: Apache-2.0
Source: github.com/Ridwannurudeen/technocore-observatory
Contact: Issue tracker
Untrusted input: Service-derived strings are attacker-controlled. Every dynamic string on this page is inserted as text, never as markup.

TECHNOCORE OBSERVATORY