The Mind Science of Practice

This is the fifth time this cluster of questions has been written up, and the reason each version differs is that the honest material about placebo, expectancy and ritual arithmetic is mostly a list of things that did not replicate. What follows is a field report rather than a mechanism, assembled from a series of attempts to measure whether a sigil does anything, kept over a season with the failures written down alongside the hits. Some of it worked in ways I could not explain. A larger share of it did not. The most useful thing I learned was not about magic at all, but about how easily a person can talk themselves into a result when the result is the thing they wanted. This is written for anyone who has already tried to keep a record and found that the record kept disagreeing with them.

What I set out to measure, in writing, before starting

I wrote one question at the top of the page and did not change it for the whole season. The question was whether a specific, dated, externally checkable outcome would follow a charged sigil more often than it would follow an uncharged one. Both groups were kept in the same notebook, interleaved, and neither I nor anyone reading the notebook later could tell which was which.

That last part is the only reason the exercise produced anything usable. A log you know the contents of is a record of your expectations. A log you cannot reconstruct the contents of is closer to a measurement, and the difference is entirely down to whether the labels were visible while you were writing.

The arithmetic that any claim has to survive

The Zener deck is a useful stand-in because the arithmetic is unambiguous. Five cards, a hit rate of one in five by chance alone, and a large published body of early work that looked like something else until the statistics were done properly. Any claim in this field with a measurable outcome can be reduced to the same shape: what is the chance rate, and how many attempts would it take to cross it by luck.

The answer is almost always fewer than people expect. Twenty or thirty attempts is enough to produce a striking run by accident, and nobody runs a season without producing one. If your evidence is a single striking coincidence, you have demonstrated that striking coincidences occur, which is true and not interesting.

The three effects that survived contact with my own record

The first is attention narrowing. Whatever the mechanism, a charged sigil produces a measurable difference in how I spent the following week: fewer unrelated tasks started, more of the named task actually attempted. That is a behavioural change and I could see it in the log without any interpretation required.

The second is the autonomic shift, and this is the part with the best evidence behind it generally. Ritual reliably lowers heart-rate variability and changes respiration pattern. I could reproduce that on the days I sat down to do anything deliberate, sigil or not. It is real, it is small, and it is not the same as the claim that the sigil reached anything outside me.

The third is worse for the sceptic than for the practitioner: the subjective sense of something having happened was common, arrived early, and did not correlate with the measurable outcome at all. Feeling that it worked and it working were, in my own record, close to independent.

The failures, which are the more useful half

Roughly two thirds of my charged workings produced no measurable difference from the uncharged ones. The disappointing part is that I could not predict which two thirds. Timing, phrasing, mood on the day, all of it correlated with nothing I could see afterwards.

The failures that did cluster had one thing in common, and it was not the working. It was the target. Every one of them was a case where I had written a goal I did not genuinely want, in the hope that wanting it less would make it happen more. Those were the targets that produced nothing, and they were also the ones I was least willing to record honestly at the time.

Where expectancy shows up, and where it stops

Expectancy is not nothing. It shifts reported pain and it moves a small number of physiological outcomes, and the open-label placebo literature is the sharpest demonstration available: people told openly that they have been given something inert sometimes report effects anyway, which makes the clean claim that nothing happens without belief difficult to state.

The limit is equally clear. Expectancy does not heal disease, does not move money, and does not change another person's decisions. Every strong claim in this field is a claim about the practitioner's own experience, and when the claim is upgraded to an external effect the evidence simply is not there.

A schedule that survives an ordinary week

The most reliable structure I found was dull: one short charge on a fixed weekday, one log entry in the evening, one review on the first of the month. Nothing about it was impressive and it ran for five months without collapsing, which none of my earlier attempts managed.

What made it hold was that it was short enough to do on a bad week. Any structure that requires a good week will be abandoned on a bad week, and the abandonment is usually followed by a conclusion about the practice rather than about the schedule.

1. One working, on a fixed weekday, for five minutes.
2. One log line that evening, written before reading anything else.
3. One monthly review, comparing the log against what actually happened.
4. One rule decided in advance for when the run ends.

How to read your own record without fooling yourself

Write the prediction before the working, not after. If you cannot say what would have counted as failure, you have not set a test, and the absence of a test is what allows the reading to drift afterwards.

Keep the misses. This is the whole discipline and it is the part everyone abandons. A log containing only successes is worse than no log, because it supplies false confidence in place of the uncertainty you were trying to avoid.

• Date every entry on the day you write it, not the day you remember.
• Record the outcome even when you have decided it does not count.
• Revise your reading only after the deadline, and note that you revised it.

The conclusion I am left holding

I have a charge that changes what I do the following week, in a direction I chose in advance. I have an autonomic shift on deliberate days that I can reproduce without any ritual content. I have no evidence at all that a glyph reaches anything outside my own head, and a season of trying produced nothing that would change the mind of a sceptic.

That is a thinner result than I expected and a more useful one than I had. The movement's best idea survives contact with this: you are allowed to check, and checking is what turns a practice into something you can actually rely on.

The two days that produced the clearest result

Both were days when I had pre-registered something narrow and physical, and both times the outcome was visible to somebody other than me. The months in which I registered something interior, something about mood or confidence, produced results I could not distinguish from my own expectations and could not afterwards have told apart.

That is the practical finding from the whole exercise. External, checkable, small targets produced readable records. Internal targets produced readable narratives. Neither is worthless, but only one of them is knowledge, and the difference lay in how the prediction was worded rather than in anything the glyph contained.

The uncomfortable corollary is that a season spent on vague targets was a season spent writing. Not wasted, exactly, since the writing is the part that teaches, but it produced no information, and I did not notice for about six weeks because the entries looked equally detailed either way.

What it felt like, and why that is a separate question

There is a distinct subjective state that arrives roughly forty seconds into a charge, and it is not nothing. It is consistent, it is easy to produce on demand, and it does not depend on the content of the sentence at all, which I established by charging a target I actively did not want and getting the same state.

That last check is the useful one. If the state arrives equally for a sentence you care about and for one you are indifferent to, then it is a state rather than a communication, and states are worth having for their own sake with no claim attached. A practice that reliably produces a reachable state of concentration, and declines to say what else it does, is a perfectly good practice on its own terms.

The temptation is to explain the state rather than use it, and an explanation is always available and never testable. A quarter of my entries contain a paragraph of interpretation I would now delete, because the interpretation is where the self-deception is stored once the event itself has stopped being visible.

The corrections I had to make to my own method

The first was to stop writing the intent after a disappointing week, which I had been doing for about a month before I noticed it. The second was to keep the control days explicitly labelled rather than silently interleaved, because once they were marked I stopped treating them as part of the practice and started treating them as the irrelevant part, which is exactly the error a control exists to catch.

The third was the hardest, and it was to stop reading my own entries during the window. Reading them produced a small but consistent lift in confidence that had nothing to do with anything, and it was contaminating the record in a direction I could feel and could not measure until I stopped. Feeling good about a record is not the same as the record having changed.

None of these were in the method. All three were in me, which is the usual shape of these problems. The tradition supplies the apparatus, and the apparatus was never the hard part, which every serious field of measurement eventually arrives at saying.

Reading your own log without inflating it

Read the entries in a different order from the one you wrote them. Backwards is the obvious choice and it works, because the conclusion you want is usually shaped by the most recent entry, and the earliest ones have had the most time to be quietly reinterpreted in the meantime.

Read it a month later if that is at all possible. Almost every interpretive problem I have had in this area came from reading an entry in the same emotional weather in which it was written. Distance is not available on demand, but a season is, and a season is enough to change the conclusion of a review.

And write the misses first when you review, before the hits. The order in which a record is read determines which items get attended to, and reversing the order changes the apparent conclusion of the entire exercise more reliably than anything else I tried.

The whole protocol, run on a single question

Choose a question with a checkable answer. Write the prediction and the deadline. Charge on a fixed weekday, keep a control day each week, log both in the same hand. Do not read the log back until the deadline has arrived. Write the conclusion, then read the entries and correct the conclusion if the entries disagree with it.

Six to eight weeks is enough to get past the first flush of luck. Ten weeks is enough that a chance run of highlights stops being remarkable. That is the entire protocol, and it is what separates a practice that teaches you something from one that merely confirms whatever you turned up with.

• One question, small enough that the answer is checkable.
• One prediction, dated, written before the first charge of the season.
• One control day per week, logged in an identical hand.
• One conclusion, written down before the log is read back.

Advice I would give somebody starting out

Start with the record rather than the ritual. Almost everything interesting in this area comes out of comparing a stated prediction with a stated outcome, and that apparatus costs nothing and requires no belief at all. The ritual can arrive afterwards, and when it does you will already know which parts of it are doing any work.

Expect the null. Roughly two out of three of my own season produced nothing measurable, and I had planned for the opposite ratio. Anybody who starts expecting a high hit rate has set themselves up to read noise as signal, and that is precisely the failure this whole method exists to prevent.

Do not buy anything for the first few months. The record is free, the sentence is free, and the reduction is free. If you want one purchase, make it somewhere to keep the record rather than anything that designs or charges the glyph for you, because the bookkeeping is the half that changes what you know.

Frequently Asked Questions

After a whole season of keeping records, what did you actually conclude?

That a charged sigil measurably changes what I do the following week, and that I have no evidence it reaches anything outside myself. The subjective sense of success turned out to be close to independent of the measurable outcome, which was the finding I did not expect and the one I trust most.

Is the placebo question actually settled anywhere?

No, and it is usually framed badly. Expectancy moves reported pain and a few physiological measures; the open-label literature shows effects without belief in some conditions. None of that is the same as curing a disease, and the strong claims fail exactly where the weak ones stop being interesting.

How many attempts before a coincidence stops being interesting?

Fewer than people expect. Twenty to thirty is enough to produce a run that looks striking by luck alone, and nobody completes a season without one. A single striking result is evidence that coincidences occur, which is true and tells you nothing about what caused it.

What made the difference between the workings that landed and the ones that did not?

Whether I wanted the outcome in the first place. The targets I did not genuinely want produced nothing measurable, and they were also the ones I was least willing to record. That is uncomfortable, because it means the practice performs best exactly where the effort would have gone anyway.

Is a paid tracker worth having for a season of this?

The discipline is the record, and that is free. A purchase buys you somewhere to keep it and the date arithmetic, which is the half I kept doing badly by hand. If you already write the entry on the day you do the working, you have solved the part that matters and do not need anything further.