itOpened Blog · Cancellation Data
Tomorrow's Openings, Finally Measured
When we split campsite survival by lead time in our last post, one bucket was missing entirely: openings for tonight and tomorrow had produced no completed life cycles, so we published the gap instead of a guess and promised to return when the sample existed. It exists now — for tomorrow, at least. The closure log holds 56 completed life cycles for stays starting the day after detection, and they lived a median 3h 35m: longer, not shorter, than openings detected in the final week.
Quick answers
- How long does an opening for tomorrow last?
- In the current closure log, openings detected one day before the stay lived a median 3h 35m across 56 completed life cycles — longer than the 2–7 day bucket's 2h 45m (n=325) and longer than the all-campground median of 2h 37m.
- What about openings for tonight — same-day stays?
- Still unmeasured, honestly: the alert log holds 8 non-Cape-Point openings detected for a stay starting that same day, but 0 of them have completed a measured life cycle in the closure log, so this post publishes no same-day survival figure in either direction.
- How many last-minute openings are there in total?
- 196 of the 1762 non-Cape-Point confirmed openings in the alert log were detected zero or one days before the stay (188 at one day, 8 at zero). Last-minute inventory is a real slice of the flow, not a rounding error.
- How was survival measured?
- Each confirmed opening is followed until two consecutive polls find it gone. Survival is the time from first confirmed detection to confirmed disappearance. Sample for this update: 675 completed life cycles for November–December 2026 stays, computed from the closure log at build time.
Watch it with itOpened
Watchman reopened 111 times in our log — including on the eve of the stay.
Tomorrow's openings live a median 3h 35m — long enough to book, short enough that you have to already be watching when they appear. Our scanners poll Watchman around the clock and text you while the site is still alive.
Proof, from our own scanners: 111 confirmed openings at Watchman since Oct 29, 2026 — each one re-verified before logging. See the live cancellation index.
Watch Watchman openingsFree to start. We never book, hold, or resell — you book it yourself, in your own account, at the normal price. Fees pay for monitoring; openings are first-come and cannot be guaranteed.
Good data work has a loose-ends problem: the honest post is the one that names what it cannot yet measure. Our lead-time post did exactly that. Splitting 619-odd completed opening life cycles by how far ahead of the stay each opening appeared, it found a stubbornly flat clock — and it flagged, in bold, the missing bucket: no completed life cycles at all for openings detected less than two days before the stay. Tonight-and-tomorrow openings existed in the alert log; none had yet lived and died inside the measurement window. We said we would publish the bucket when the sample existed. The scanners kept running. This is that publication.
The updated table
The closure log now holds 675 completed life cycles. Split the same way as before — lead time is the gap, in Eastern calendar days, between the day an opening was detected and the first night of the stay — with the new one-day bucket added:
| Opening appeared | Completed life cycles (n) | Median survival |
|---|---|---|
| 1 day ahead (tomorrow) | 56 | 3h 35m |
| 2–7 days ahead | 325 | 2h 45m |
| 8–14 days ahead | 181 | 2h 00m |
| 15+ days ahead | 113 | 2h 50m |
The new row is the surprise. Openings for tomorrow lived a median 3h 35m — the slowest bucket in the table, slower than the 2–7 day heartland at 2h 45m and slower than the overall median of 2h 37m. The intuition that last-minute openings must be the most frantic, shortest-lived inventory in the system is not just unsupported by this sample; the sample points the other way.
Why might that be? Two mechanisms are consistent with the numbers, and we label both as interpretation, not measurement. First, competition thins at the last minute: a stay starting tomorrow can only be taken by someone who can actually go tomorrow — gear ready, time off already in hand, drive distance short. The universal queue that devours a three-night block two weeks out is, at one day's notice, a much smaller room. Second, supply at one day out is disproportionately deadline-driven: holders who finally fold on the eve of a trip release sites they had genuinely intended to use, and those sites arrive singly, at odd hours, into that thinner room. Either way, the practical translation is the same, and it is genuinely good news for flexible campers: tomorrow's openings are not a faster game. They are, on this evidence, a slightly slower one.
What did not change
The rest of the table kept its shape as the sample roughly doubled-and-more. The clock across the measurable lead times remains a band, not a cliff: every bucket from one day to fifteen-plus sits between 2h 00m and 3h 35m at the median. There is still no lead time at which an opening lives long enough for a once-a-day check to be a plan, and now there is direct evidence that there is no lead time — tomorrow included — at which openings die so fast that only luck explains a catch. The machine-versus-human gap is a detection gap, at every distance we can measure.
And what is still missing: tonight
One bucket remains empty, and it stays labeled empty. Openings detected for a stay starting that same day do occur — 8 of them in the non-Cape-Point alert log so far — but none has completed a full measured life cycle in the closure log: a same-day opening that is still bookable when its stay date arrives ages out of the measurement instead of dying inside it, and openings detected and re-booked between polls can leave no complete cycle to measure. So the honest sentence is unchanged for same-day stays: we do not know their median survival, we will not estimate it from the tomorrow bucket, and when a method that can measure it exists, it will get its own post. A missing number, stated, is worth more than a plausible number, printed.
How to hunt tomorrow
Treat tomorrow as a first-class target, not a consolation prize. With 196 zero-to-one-day openings in the alert log and a 3h 35m median life on the measured ones, the eve-of-trip market is deep enough and slow enough to be a deliberate strategy for anyone within a short drive of a watched campground. Be packed before the alert. The constraint on a tomorrow opening is not the clock — it is you. The hunters who can use a 3h 35m window are the ones whose answer is a booking screen, not a packing list. Keep the same response discipline as any other bucket. A slower median is not a slow opening: half of tomorrow's openings still died inside 3h 35m, and the distribution underneath every median on this page includes openings that lived minutes. Watch broadly on short horizons. Tomorrow inventory is single-site and scattered; a portfolio of acceptable campgrounds converts scattered singles into a probable trip.
The limits, plainly
The one-day bucket rests on 56 completed life cycles — a real sample, printed with its n, and still young enough that the median will move as the log grows. Right-censoring applies as always: openings still alive at sample close are excluded, biasing medians modestly downward. The season is shoulder season — November and December stays, detected in late October and the first hours of November — and a summer sample could compress every row. Facility mix is not controlled for: the tomorrow bucket's campgrounds are whoever cancelled on the eve, not a matched panel. The live counts are on the cancellation index, the method is on the methodology page, and the prior post stands as the record of what we said when this bucket was empty: nothing. That is the deal this site makes — the number appears when the measurement does, not before.
Watch it with itOpened
Watchman reopened 111 times in our log — including on the eve of the stay.
Tomorrow's openings live a median 3h 35m — long enough to book, short enough that you have to already be watching when they appear. Our scanners poll Watchman around the clock and text you while the site is still alive.
Proof, from our own scanners: 111 confirmed openings at Watchman since Oct 29, 2026 — each one re-verified before logging. See the live cancellation index.
Watch Watchman openingsFree to start. We never book, hold, or resell — you book it yourself, in your own account, at the normal price. Fees pay for monitoring; openings are first-come and cannot be guaranteed.
Frequently asked questions
How long does a campsite opening for tomorrow stay open?
A median of 3h 35m, across 56 completed opening life cycles in our closure log — openings detected the day before the stay. That is longer than the 2–7 day bucket's median of 2h 45m.
Do last-minute campsite openings vanish faster?
Not on the evidence now available. The one-day bucket is the slowest in the updated lead-time table. The earlier finding — a broadly flat clock across lead times — holds with the last-minute bucket now filled in for tomorrow. Same-day survival remains unmeasured.
Why is there still no figure for same-day openings?
The alert log contains 8 non-Cape-Point openings detected for a stay starting that day, but 0 completed a measured life cycle in the closure log — a same-day opening can age out at its stay date instead of being observed dying. We publish no figure rather than an estimate.
How many campsite openings appear at the last minute?
196 of the 1762 non-Cape-Point confirmed openings in our alert log were detected zero or one days before the stay, out of 2161 confirmed openings in total (Cape Point Campground's bulk seasonal event excluded, as in every analysis here).
Sources
- itOpened scanner closure log (engine/closures.jsonl), sample closing Nov 01, 05:30 AM ET: 675 completed opening life cycles (ambiguous records excluded). Lead time = stay start date minus the Eastern date of the opening's alert; survival = first confirmed detection to confirmed disappearance (two consecutive gone polls). Medians by lead bucket: 1 day ahead (tomorrow) 3h 35m (n=56); 2–7 days ahead 2h 45m (n=325); 8–14 days ahead 2h 00m (n=181); 15+ days ahead 2h 50m (n=113). Completed life cycles with a lead of 0 days: 0. Overall median survival 2h 37m. All figures computed from the log at build time.
- itOpened production scanner alert log (engine/alerts.jsonl), Oct 29, 10:54 PM ET – Nov 01, 05:30 AM ET: 2161 confirmed openings in total, 1762 excluding Cape Point Campground's 399 bulk alerts; 196 of the non-Cape-Point openings were detected 0–1 days before the stay (188 at 1 day, 8 at 0 days).
- itOpened, 'Do last-minute campsite openings vanish faster?' (/blog/do-last-minute-campsite-openings-vanish-faster/): the prior lead-time split, which reported no completed life cycles under a 2-day lead and promised this follow-up when the sample existed.
- itOpened methodology (/methodology/): poll cadence, confirmation standard, and closure definition.
Every figure in this article traces to one of the sources above. If a number isn't sourced, we don't print it.
Last updated 2026-10-30.