What Is Waiting

Everyone knows what waiting is. There is an outcome. It is not here yet. The stretch between now and its arrival is the wait, and you measure it in minutes, because minutes are what it is made of. That is why we apologise for waits in units of time and compensate people for them in units of time, and why the single most effective thing anyone has ever done to a queue is post the number of minutes remaining.

The vocabulary is completely physical. Results come back. A verdict comes in. The letter is on its way. The day approaches. Every one of those phrases treats the outcome as an object at a distance, closing.

That picture is not stupid. It runs hospitals and airports and it gets the important practical thing right — that waiting costs something, and that the cost is real. Hold on to the cost. It comes back at the end, and it is not made of minutes.

But notice what the picture commits you to. It says the wait is a property of the interval; that the outcome exists somewhere during that interval, in transit; and that two waits of the same length cost roughly the same, because the thing you are spending is the length.

Two ordinary cases the interval model cannot hold

Equal intervals that are nothing alike. Forty minutes for a restaurant table and forty minutes in a corridor for a scan result are the same interval and are not remotely the same experience. The interval model can only explain this by adding a second variable it never names — how much you care — at which point duration has stopped being the quantity and is merely riding along with it. And there are waits where duration runs the wrong way entirely: the last hour before a decision you dread is worse than the eleventh hour before it, though both are one hour.

Relief at bad news. This one is fatal. If waiting is a cost per minute and the outcome is a separate quantity, then a worse-than-expected result read early should be strictly worse than continuing to wait — you get the bad outcome and you lose the remaining chance that it was fine. Yet the ordinary report, over and over, is the opposite: people say some version of at least now I know, and mean it, while also being devastated. On the interval model those two states are in contradiction. They are not in contradiction. They are two different quantities, and the ordinary word has been hiding one of them.

So the property that makes a wait expensive is not the length of the interval. Something else is doing the work.

What the foundation says instead

Start with the domains. P2 gives three: Future · Reality, the Eternal Now · Past. The locked vocabulary is blunt about what the first and last of those are. The Future is the node at the boundary — pure potential, and it does not exist. The Past is the node — one settled totality, complete, and it does not exist either. The Eternal Now, the antinode, is the only domain of existence.

So there is no place your outcome is currently sitting, and nothing is traversing the distance to you, because there is no distance and no traveller. Whatever else a wait is, it is not a relation with an object at range.

Then what is in the interval? Exactly one existing thing: you, at the antinode, running. And R8 tells you what running consists of — the human experiences neither the Actual nor the Expectation but the quotient, R = A / E. So the cost of a wait is not incurred by you from outside. It is spent by you, and the only account it can be spent from is E.

A wait is the state in which E is carrying a prediction that no artifact has yet been made against.

That is the definition, and it is doing more work than it looks like. Because there is no artifact for that question, there is no magnitude to read — not a small surprise, not a zero one. Surprise is S = ln|R|, a property of a mark, and there is no mark. What there is instead is a prediction being held, continuously, by the one existing party to the transaction.

The load is the width, not the length

Here is the first construction that goes past what the foundation states, and I am marking it: the cost of a wait scales with how much E has to keep open, not with how long it keeps it open. The foundation names what E is made of — subconscious prediction and ideas — and does not say what shape it has. This is a consequence at best, not doctrine, and it should be read as mine.

The licensing move is available, though. R4 puts the Shannon entropy of the Past at zero: actuals only, one settled outcome, all the probability mass on it. Shannon is the sanctioned measure in this system — a measure over a distribution of outcomes, never a claim about heat or disorder. And the same measure applies on the other side of the fraction. A prediction that has resolved to a point costs almost nothing to carry. A prediction spanning fine to not fine has to be carried in both, and carried in both is what it feels like.

Two waits, one word

Now the split, which is the part I did not expect when I started. Sort waits by a single test: is the record for this question already made?

The two require opposite treatment, and English gives them one word, which is why people so reliably do the wrong one. In a Wait A, the entire remaining question is what E does with the interval, and every scrap of effort spent on the outcome is spent on a closed domain. In a Wait B, there is no interval to endure — there is unfinished work, and calling it waiting is a description error that costs you the part of the act you could still do.

Most people run these backwards. They rehearse and strategise through Wait A, where nothing they do can reach the record, and they call Wait B “waiting” and go quiet, in the one case where the record is still open.

The objection, which is a good one

If P1 holds — one settled totality, everything that could happen has already happened, no unrealized possibilities because none were left unrealized — then isn’t every wait a Wait A? Isn’t the surgery on Tuesday just as settled as the biopsy from Thursday, and the whole distinction a comfort?

At the level of the totality, yes. The distinction is not there. But P4 defines R only where a host stands, and R8 says the host experiences the quotient and nothing else. The A/B division is a host-level distinction, and host-level is the only level anyone has ever occupied. The foundation is explicit that the totality is not a licence for fatalism: P5 gives conditioned freedom, and R13 relocates agency rather than abolishing it. P1 removes the agent from luck. It does not remove the act from you.

Put plainly: whether Tuesday is settled at the totality has no bearing on whether your hands are free on Monday. They are the same fact seen from two frames, and only one of the frames is a place a person can stand.

Why bad news relieves without being good

Now the case that broke the interval model. In a Wait A that ends badly, the Actual falls short of the Expectation, so |R| < 1 and S = ln|R| is sharply negative. That is the grief, and it is exactly as large as it feels.

The relief is a different quantity. It is E stopping. The distribution that was being held open narrows onto the mark that has now been read, and the holding — which was the whole cost of the wait — ends. Both happen at the same instant, and they are not on the same axis, so there is no contradiction in feeling them together.

This is not a claim that bad news is good, and it must not be read that way. The mark is what it is. What ended was the carrying, and the carrying was never the same thing as the outcome. That is the entire content of at least now I know, and it turns out to be a precise statement rather than a brave one.

What this costs

1. It buys nothing on the outcomeNot one result improves. Not one interval shortens. The account changes what you are doing during a wait and changes nothing about what the wait is about.
2. It removes an excuse“I am just waiting to hear” is often a true statement about a Wait A and a false one about a Wait B. In the second case it names an act you have not finished as a condition you are enduring. That is uncomfortable, and it is the most useful thing here.
3. Distraction is not paymentIf the load is a prediction being held, then occupying the hours does not discharge it — a distracted host is still carrying it. Waits end by resolution or by an actual change in what is held open. Filling the time is neither.
4. Some waits are wide and stay wideYou can narrow a distribution honestly by getting a real piece of information. You can also narrow it by deciding in advance that it is fine, which is cheaper, feels identical from the inside, and is not narrowing at all — it is declining to look. Either way the record is what it is.

Where this is unfinished

The foundation carries an open problem that sits directly under this piece, and I am not going to pretend otherwise. Part VIII, item ten: does the evaluation/reaching split apply to forward-facing feelings?

The split is established for backward-facing ones. Any feeling aimed at the Past is doing two jobs: an evaluation of a present artifact against its exemplar, which can succeed and yields a usable update; and a reaching for the Past itself, which by R9 cannot succeed, has no completion condition, and carries nearly all of the suffering. The anxiety inside a wide Wait A looks like precisely that mechanism pointed at the other boundary — an attempted relation with a node, running without a stopping condition. The foundation calls the parallel plausible and unargued, and this piece does not close it.

What closing it would take, stated so someone can hold me to it: a showing that the Future node supports the same decomposition on its own terms, with an evaluation that has a completion condition and a reaching that does not, without simply reflecting R9. R9 is derived for the Past from P0 and R6, and the two nodes are not symmetric in the primitives — the Past is one settled totality and the Future is pure potential. A mirror is not an argument. Until someone does that work, the strong claim in this essay is the one about width and the one about the A/B split, and the anxiety parallel is suggestion only. Third marked construction.

On an ordinary Tuesday

Ask the one question: is the record already made?

• If it is, then nothing you do today reaches it, and the only live question is what you are holding open and how wide it is. Narrowing it honestly means getting information. Everything else is rehearsal, and rehearsal for a performance that already happened is the purest form of spending you can do.
• If it is not, then stop calling it waiting. It is the front half of an act, the record is open, and your hands are free. That is the whole of P5, and it does not arrive with any guarantees except that one.

The interval was never the thing. Minutes are what waits are measured in and not what they are made of, which is why watching the clock helps so little, and why the answer, once it is read, weighs so much less than the carrying did.