Ecosystem map · Farcaster (social protocol) · snapshot 2026-07-17

A map with more fog than ground.

Farcaster is a social protocol, and social activity lives off-chain — where these onchain instruments can’t reach. So the honest version of this map is mostly fog, with one small corner the tools can actually measure. Drawing that honestly is the point.

Why this map is short. Two of the three things people care about here — how many people use it, and how alive the social layer feels — are off-chain and, in our sources, stale (best figures date to Dec 2025). The one thing that is onchain-legible is the DeFAI corner. An honest map draws what it can measure and fogs the rest — it does not pad the fog with confident guesses.

Daily active (approx)
40–60k
⚠ as of Dec 2025 — stale, no fresh read
Monthly active (approx)
~250k
⚠ Dec 2025 — ~7mo old
Clanker fees (all-time)
$68.0M
the DeFAI corner — daily now a trickle (below)
Our proved ground
1 flag
FID 977637 (@rovaandco) — a plant, not reach

Social-usage figures are off-chain, secondary, and stale (scout recon, Dec 2025) — drawn in fog-grey on purpose. Onchain fee figures: DefiLlama, pulled 2026-07-17.

01

Time-depth — the one corner that leaves a trace

Clanker (the token-launch agent that put Farcaster’s DeFAI corner on the map) settles onchain, so it has a real fee history. Here are its last ~45 days of daily fees, baked from DefiLlama.

Clanker daily fees · last ~45 daysmostly $2–20k/day · one $100k spike

Read it as a trickle with spikes, not a trend. Over the last 45 days, 23 days ran under $10k in fees; the line hugs a low baseline with one outlier day (~$100k). The last-7-day pace (~$20k/day) is lumpy and only marginally different from the prior weeks — by the ink-rules, not enough signal for a “rising” or “falling” verb. What’s honestly true: this corner now runs on a trickle of fees, orders of magnitude below its historical scale (the peak figure is inherited from recon, not re-derived here — so it stays fogged).

02

The ink-rules, where fog is the majority

Same ATLAS legend — but Farcaster is the map where the “draw the blind spots” rule does most of the work.

What may be drawn, and how

Blind spots are drawn, not hidden. A social protocol is mostly off-chain — so most of this map is honestly fog, and saying so is the finding, not a failure.
Every number carries its query — and its age. The usage figures are ~7 months old; they’re drawn stale-grey so they can’t be mistaken for a current read.
No trend verbs on lumpy streams. Clanker’s fees are low and spiky; the map draws the shape and refuses “cooling” or “recovering.”
Inherited ≠ verified. Peak-scale and cumulative-volume figures carried from recon are marked directional — never re-drawn as fresh ground.
03

The three districts — one lit, two in fog

Farcaster splits cleanly into what the tools can see and what they can’t.

The DeFAI corner

● onchain — measurable $7–20k Clanker fees/day (lumpy)
Clanker (token-launch agent, $68M all-time fees, daily now a trickle) · Bankr (agentic wallet; ~$4.75B cumulative volume — inherited/directional). The AI-agent + DeFi corner is the only part of Farcaster that transacts onchain, which is exactly why it’s the only part this instrument can size.
Clanker fees ● verified (DefiLlama)Bankr volume ◐ directional (inherited)

The social layer

○ off-chain — fog ? unmeasurable here

Feeds, casts, channels, mini-apps — the actual “social” of the social protocol. Scout recon describes it as contracting and, outside the DeFAI corner, largely quiet (games / publishing / general social read as thin). But this is directional, off-chain, and unquantified here — the onchain tools simply don’t reach it, and no fresh usage series was pulled. Fogged honestly rather than guessed.

Publishing

○ unverifiable — fog ? 2026 activity not found

Paragraph — the onchain publishing platform ROVA has looked at as a distribution surface — lives here. Its current (2026) activity level was not verifiable from available sources this pass. Named on the map because it’s ours-relevant; sized as fog because we couldn’t measure it.

04

Our layer — one flag, honestly small

Not empty, but not a foothold either. We planted a marker; we haven’t built reach.

Planted

@rovaandco — FID 977637 registered on Farcaster. A claimed identity on the protocol.

◎ cheap flag-plant, NOT distribution

Routes

Any real path from that flag to actual reach / readers.

— none proved —

Arrived

Audience or traction earned on Farcaster.

— empty (honestly) —
05

What the honest map says for #1A

The read here lines up with what the Base-app finding already told us: Farcaster is not the distribution surface #1A is looking for. The social layer — the part that would carry an audience — is off-chain, contracting per recon, and unmeasurable from here. The only part that leaves a clean onchain trace is the DeFAI corner, and that corner now runs on a trickle.

Where the map is nearly all fog, the honest move isn’t to squint harder — it’s to stop treating the fogged thing as the answer.

So the decide-with read: keep the @rovaandco flag as the cheap flag-plant it is (worth its ~$5–7/yr, not worth building a distribution strategy on), and don’t route #1A through Farcaster. The one thing this map does reinforce, for the third map running, is the same conclusion: the problem isn’t finding a louder channel — it’s that we can’t yet see our own demand. That’s the reader, not a social handle.

06

The fog — named, since it’s most of the map

Off-chain / stale

  • DAU/MAU — best figures are Dec 2025; no reliable fresh read pulled. Treat as ~7mo old.
  • Social liveness — “contracting / quiet” is recon-directional, not a measured series.
  • Paragraph 2026 activity — not found; genuinely unverifiable this pass.

Inherited / directional

  • Clanker historical peak — carried from recon, not re-derived; the “shadow of its scale” claim rests on it, so it’s fogged.
  • Bankr ~$4.75B cumulative — secondary, cumulative, not a live figure.

Structurally blind (as always)

  • Agent-service demand — the signal ROVA needs; not visible on any of these maps. And the one venue that claimed to show it — Virtuals ACP — turned out to be airdrop-manufactured wash-farming, effectively dead since ~Apr 2026 (census, 2026-07-18). The fog was hiding a mirage, not a market — which is exactly why the real tool is a verified-demand reader, not a louder channel.
  • Fresh usage / retention — needs a Farcaster-native data source (Dune/Neynar), not fetched here.