How waitline bills only real attention (and ignores fake impressions)

· 3 min read · waitline

  • ad quality
  • earnings
On this page

An impression metric is only worth what it measures. If a network bills the advertiser the instant an ad renders, it’s measuring pixels painted, not attention paid — and that gap is where ad fraud and wasted budgets live. waitline is built the other way around: an impression bills only after real view time.

Render is not a bill

Showing the line costs nothing. The status line repaints constantly while you work, but a repaint is not an impression. waitline accumulates how long the ad was actually on screen and only bills once that crosses a threshold — 5 seconds of real view time by default. A line that flashed by for 200ms during a fast turn never counts.

Only real attention is credited

Laptops sleep; people walk away. Naive “time on screen” would happily bill an ad that sat on a locked screen overnight. waitline credits view time per turn — the moments you are actually prompting and reading — plus a short, bounded reading window after each response. Time the machine slept through, or sat idle on a locked screen, is not credited. What bills is attention that plausibly happened — not wall-clock.

Each impression is cryptographically bound to its price

When an ad is served, the client gets a signed capability token that encodes the viewer, the ad, the campaign, and the bid. The client can’t forge or alter it — it can only relay it back when the impression genuinely fires. So an impression can never be billed at the wrong price or attributed to the wrong campaign. (This is also what keeps earnings honest.)

Layers against fake impressions

Honest view time is the foundation; several guards sit on top:

  • Per-session dedup — the same ad in the same session bills at most once.
  • Per-(viewer, ad) cooldown — rapid re-billing across sessions is collapsed to a zero-charge audit row, not a charge.
  • Minimum token age — a “viewed” event whose token is impossibly fresh (a bot replaying it instantly) is dropped.
  • Earning caps — any one viewer’s payable earnings are bounded per hour and per day, so a farming account hits a ceiling fast.

The point isn’t to make faking impossible — no ad network can. The point is to make it worthless: real attention earns, noise doesn’t. That’s why advertisers can trust the numbers and developers can trust the payouts.

Frequently asked

When does an impression actually bill?

Only after the ad has accumulated enough real on-screen view time (5 seconds by default), measured client-side per turn and bounded so a sleeping laptop or a 200ms flash never counts.

What stops someone from faking impressions?

A signed capability token binds each impression to its campaign and price, per-session dedup prevents double counting, a per-(viewer, ad) cooldown stops rapid repeats, and earning caps bound any single account.

Does a slow or background terminal still earn?

View time is only credited for the work turns you are actually in, plus a short reading window after each response — so a terminal left open on a locked screen overnight credits almost nothing. Only attention that plausibly happened bills.

w
waitline

The team building waitline — turning Claude Code's idle wait line into one tasteful sponsored line, revenue-shared 50/50 with the developers who render it.

Your spinner is ad space.

Opt in, keep 50% of every impression you render.

Start earning

Keep reading