Gnosis, Belief and Paradigm Shift

Almost every disagreement about magic in this field runs through a single word, and the word is belief. Not belief in the sense of faith, but belief in the technical sense: the stance you take toward a proposition before evidence has arrived. The argument below is arranged around that, because getting the technical sense right is what allows somebody to be fully committed and fully sceptical at the same time, which is the position most people who actually practise here end up in and almost nobody describes clearly. It covers what the parabola commits you to, how confirmation bias operates as a design constraint rather than a reproach, and what a record built to disagree with you actually looks like in practice.

Belief as a technical term, not a moral one

The word is usually used to mean something vaguer, which makes it useless for argument. In its technical sense it means the credence assigned to a proposition before the relevant evidence arrives, and that quantity can be held consciously, revised, and reported on.

This is why the strong versions of the sceptic's position fail. They equate practising with believing, which conflates a provisional commitment with a closed one, and they then never have to engage with the actual results. It is also why the believer's position is more modest than it sounds: holding a sincere working commitment while keeping a record is not a contradiction.

The parabola and what it does to the argument

Taken by itself, the parabola holds that a working has no built-in morality at all, and that whatever character it ends up with is borrowed from the aim you brought to it. That idea did a great deal of the practical work during the seventies and it is still the most useful thing the tradition produced for anyone nervous about scope.

The cost is real. If nothing has an inherent character, nothing is excluded by that fact, and the only limit is your own judgement, which is a less reliable instrument than a rule. The parabola hands over the responsibility and then leaves you carrying it, and that is usually a fair trade for people who would rather not be governed by someone else's list.

What commitment buys, and what it does not

Commitment buys one specific thing: you will do the steps. Write the sentence, reduce it, charge it, log it, sit down on the appointed day. Most of what gets attributed to belief in this field is actually the by-product of doing those things repeatedly.

It does not buy accuracy of interpretation, and the record is what protects you from your own reading. A committed practitioner without a log has more access to false confirmation than a sceptic with one, because the sceptic is not generating interpretations in the first place.

• Commitment produces the record. It does not interpret the record.
• The steps are the mechanism, not the belief behind them.
• A log plus scepticism is worth more than commitment plus none.

A record built to argue with you

The effect is thoroughly documented and no reflection of character. It is what a forecasting arrangement does whenever the rule handed to it can be met by nearly any event. A working counts as successful once the goal is reached, unsuccessful once the goal is dropped, and beside the point when the target was mistaken regardless, so the rule is met by the world no matter what the practice did.

Which makes it a design problem. The remedy is a narrow target, a pre-registered prediction, a fixed deadline and a record containing the misses. All four cost nothing and all four are abandoned by nearly everybody, which is the reason this field looks the way it does from outside.

Expectancy, with the edges marked

Expectancy moves reported pain measurably and touches a handful of physiological measures. That is a real and bounded effect, and the open-label placebo literature is the cleanest demonstration: people informed openly that the substance does nothing sometimes report effects regardless, which makes the claim that nothing happens without belief hard to state in its strong form.

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

A Bayesian practice, and why it is not a betrayal

Keeping a record, pre-registering a prediction and updating when the evidence comes in is a Bayesian approach to practice, and it has nothing to do with scepticism. You can believe a working will produce something and still be willing to revise when it does not, and in fact that is the only version of belief that produces any information.

The practical version is dull. Write what you expect. Write the date by which you will check. Write what happened. Notice the direction of the result, not its size. Repeat. Over a season this produces a rough estimate of how often your workings land, which is a number very few people in this field have ever calculated about themselves.

Building the record so it can argue back

Four fields are enough, and the fourth is the one everybody leaves out. Date, prediction, deadline, outcome. The outcome field must be able to hold a miss in the same sentence as a hit, without qualification.

It should be written in the same session, in the same tone, with no narrative. An entry rewritten afterwards into a tidy narrative is worse than leaving the page blank, because it hands you certainty where the page was supposed to hold a question.

1. Date the prediction before the working, not after.
2. Fix the deadline before the working, not at review time.
3. Record the outcome on the day, including what does not fit.
4. Revise your reading only after the deadline, and note that you did.

Where sincerity and credulity part company

They part company at the moment the conclusion is allowed to arrive first. A credulous reading is one where the desired result is treated as established by a coincidence, and it is usually sincere, which is what makes it difficult to spot from the inside.

The test is narrow and practical. Ask what the result would have looked like if it had gone the other way, and whether you would have written that down. If there was no version of the outcome that would have counted as a failure, you did not run a test, and no amount of sincerity in the rest of the practice compensates for that.

What to do with all this on a Tuesday

One question with a checkable answer. One prediction, dated. One short working. One line in the notebook the same evening. One review on the fixed date, reading the entries only afterwards and revising the conclusion if they disagree with it.

That is the entire apparatus, it takes about twenty minutes, and it is enough to make any subsequent argument about belief considerably better informed. Everything else in this field is elaboration on those five steps, and most of the elaboration makes it worse rather than better.

Belief as a number rather than a position

The most useful move in the whole area is to treat your credence in a particular working as a quantity you could write down and revise, rather than as an identity you have. Almost nobody does this, including people who are careful about everything else.

Try it once. Before a working, write how likely you think the outcome is. After the deadline, write it again. Most people find the first number was much higher than the evidence warrants and the second much higher than the first, and finding that out once is usually more persuasive than any argument anybody can make.

• What is my credence in this specific outcome, as a percentage?
• What would move it by ten points in either direction?
• Who else would have to agree before I changed it?

Where the strong sceptic position fails

The confident version argues that practising requires believing, which is false in the technical sense and confuses a provisional commitment with a closed position. Having made that move, the position never has to look at results at all, which is why it is so common and so rarely examined.

The opposite failure belongs to some practitioners, who treat a provisional working commitment as a permanent one and then read everything afterwards as confirmation. Both errors have the same shape: a decision about how to interpret a result is made before the result exists, which is precisely what a test is supposed to prevent.

The season, laid out as a schedule

Four months, one question at a time, nothing larger than a single afternoon's work. Each block is four weekly workings with an interleaved uncharged control, a written prediction, a fixed deadline, and one review at the end of the block.

The reviews are the whole point. Read the entries before writing the conclusion and you will get a different and slightly worse answer, and doing both in that order for one block is enough to show anybody what the record is actually saying.

What the field has never done

There is no published body of pre-registered trials in this area. Not a small one, not a bad one: none. Practitioners run workings, sometimes keep notes, and almost never publish the notes, which means the entire accumulated experience of the field exists as anecdote distributed informally and never aggregated.

That is fixable at the level of one person, which is the level this page is written for. It is not fixable by argument, because there is no body of evidence to argue from. A season of your own careful record would add more to what is known than any amount of further reading, and it costs about twenty minutes a week.

The practice that produces no information and is still worth doing

Receptive attention on a narrow, checkable target is dull and it works. Ninety seconds of sustained contact with a habitual reaction, counted and written down, produces a number about how often that reaction fires. That number is small, boring, entirely checkable, and available to nobody but you.

It is also the only part of this area that a sceptic would not dispute, and it is not the part most people are looking for. Whatever else you take from a practice in this field, the trainable attention is the thing that survives contact with anybody else.

A review that can actually be defended

Read the entries before writing the conclusion and you will get a different and slightly worse answer. Do both in that order for one block and the gap between them is the honest measure of how much interpretation is doing, which is the number very few people in this field have ever computed about themselves.

The whole schedule fits in twenty minutes a week. That is the part that should make you suspicious of anybody selling you a more elaborate apparatus: if the record cannot be kept in twenty minutes, it will not be kept, and a record that is not kept is not a record.

Frequently Asked Questions

Does holding a belief make results worse?

It can, and the mechanism is specific rather than mysterious. A belief about a particular outcome makes it easier to read any event as confirmation, and the result is a record that has stopped being able to disagree with you. The fix is not less commitment, it is a prediction narrow enough that failure would be visible, written down before the working.

Is a Bayesian approach compatible with taking the practice seriously?

Yes, and it is more compatible than the alternative. A sincere commitment does not require suspending ordinary evaluation, and a practice that punishes you for checking it is asking for a kind of faith its own founders did not have. The combination most people who practise for years end up with is full commitment with an honest record.

What does the parabola actually commit me to?

To the view that a working has no character of its own beyond the aim you gave it, which means the moral question is settled by the aim rather than by the technique. The practical cost is that the only limit left is your own judgement, so the tradition cannot supply the boundary and you have to supply it yourself.

How do I know whether my record is honest?

Apply the failure test. Ask what the result would have looked like had it gone the other way, and whether you would have written that down. If nothing would have counted as failure, you were not running a test. If you edited an entry after the fact, the record has become a narrative and the two are not the same instrument.

Is there anything worth paying for in a practice like this?

The practice and the record are free and should stay that way. A purchase buys the date arithmetic and somewhere durable to keep the log, which is the least interesting half, and if you keep one line a week in an ordinary notebook then nothing on sale improves it. The one exception is a timer for a receptive-attention exercise, where you genuinely cannot count duration accurately without one.