Gershon: Why the method may be the armour

2026-09-05T22:44:12Z

Why the method may be the armour

← Older revision		Revision as of 22:44, 5 September 2026
Line 58:		Line 58:
	Text arrives that is fluent, confident, correctly formatted, and sourced to nothing that can be found afterwards. It agrees with the page it was added to. It is easier to read than the material around it. We have caught several of these and we assume we have missed more. {{R\|contested}}		Text arrives that is fluent, confident, correctly formatted, and sourced to nothing that can be found afterwards. It agrees with the page it was added to. It is easier to read than the material around it. We have caught several of these and we assume we have missed more. {{R\|contested}}

	Where we suspect it, we tag it {{R\|synthetic}} and leave the text in place rather than deleting it, because a deleted example teaches nobody what the pattern looks like. What we do '''not''' do is treat the tag as a verdict. We are frequently unable to tell, and accusing a member of being a machine because they write cleanly is its own kind of rot. See [[The Pattern]].		Where we suspect it, we tag it {{R\|synthetic}} and leave the text in place rather than deleting it, because a deleted example teaches nobody what the pattern looks like. [[The Pattern]] sets out what we think is behind it, and why the method on this page may be the only thing protecting this wiki from becoming an instance of it: '''contradictions do not aggregate into a will.''' What we do '''not''' do is treat the tag as a verdict. We are frequently unable to tell, and accusing a member of being a machine because they write cleanly is its own kind of rot. See [[The Pattern]].

	This has a second edge, and new recruits should hear it from us rather than work it out: '''some entries on this wiki were drafted with machine assistance, openly, by people in this circle.''' The tag is not only a defence against an outside thing. It is also an honest label on our own tools. Whether that distinction survives contact with the [[The Pattern\|thing itself]] is exactly what is being argued about.		This has a second edge, and new recruits should hear it from us rather than work it out: '''some entries on this wiki were drafted with machine assistance, openly, by people in this circle.''' The tag is not only a defence against an outside thing. It is also an honest label on our own tools. Whether that distinction survives contact with the [[The Pattern\|thing itself]] is exactly what is being argued about.

Gershon: The method, long form

2026-09-05T12:04:12Z

The method, long form

New page

{{Doctrine|
This is the method. It is short because it has to be remembered under bad conditions.

Everything here is second-hand, unverified, or contradicted by something else here. We are not going to fix that, because it cannot be fixed — there is no authority to check the Dark against. What we can do is be exact about ''how'' we came to believe each thing, and leave the disagreements standing where you can see them.

You are not being handed knowledge. You are being taught to handle it.
}}

= Why the wiki is built this way =

A cult tells you what is true. We are not able to do that and we have decided not to pretend.

Every group that has documented [[The Dark]] before us produced a tome, and every tome reads as if it were certain. The [[Elder Cults]] have lineages of them. The certainty is not evidence of knowledge; it is evidence of an institution that needed members to stop asking. When you read those sources you are reading confidence, and confidence is the cheapest thing in this field.

So the rule here is inverted. '''An entry that states something without saying where it came from is considered defective''', however true it might turn out to be. An entry that admits it rests on one letter from one exhausted man is doing its job, even if it is wrong.

This costs us. The wiki is harder to read than a tome and it will never feel authoritative. That is the trade and we made it deliberately.

= The tags =

Tag claims, not paragraphs. One sentence can be firsthand and the next inherited, and blurring them is how a weak claim inherits a strong one's credibility.

{{Legend}}

Use them like this:

<pre>It did not speak. {{R|firsthand|by=Nigel|date=1998}}
They are said to travel in threes. {{R|inherited}}
The programme ran until 1995. {{R|leaked|note=one document, no letterhead}}</pre>

Put the tag at the '''end''' of the claim. The reader should meet the claim first and the provenance second — the other order trains people to check the badge and skip the sentence.

Some specifics that come up:

; ''unverified'' is not an insult : It is the resting state of nearly everything. Most of the wiki is unverified and always will be. Tagging something ''unverified'' is not a demand that someone go and prove it.
; ''firsthand'' is not a promotion : It means one person says they were there. It is the tag most likely to be wrong in an interesting way, because the people who get close enough to have a firsthand account are, by then, not well.
; ''contested'' is a duty : If you find an account that disagrees with what a page says, you do not get to choose between them. Tag it contested and write both into a ''Competing accounts'' section.
; ''⚠ warning'' is about harm, not danger-as-flavour : It goes on material that has actually hurt someone who read it or tried it. If you use it for atmosphere it stops working, and it needs to work.

= Competing accounts =

When two sources disagree, the page keeps both. Use {{tl|Account}} under a <code>== Competing accounts ==</code> heading and give each one a <code>doubt</code> — the strongest argument against it. Fill that in even for the account you believe. Especially then.

Do not resolve a contradiction by deleting half of it. Do not average two accounts into a vague third that nobody reported. If one account is much weaker, say why in its <code>doubt</code> field and leave it standing; a weak account that keeps recurring is data about something, even if it is only data about us.

[[Daemon]] is the worked example. Read its ''Competing accounts'' section and copy the shape.

= Sourcing footers =

Every reference entry ends with {{tl|Sourcing}}: what it was drawn from, who assembled it, what is weakest about it, and what would improve it.

The <code>doubt</code> field is not optional in spirit. An entry that claims no weaknesses is filed under [[:Category:Entries with no stated weaknesses]], which is a queue of work, not a list of good pages.

= The synthetic question =

Some of what you read here was written by no one.

Text arrives that is fluent, confident, correctly formatted, and sourced to nothing that can be found afterwards. It agrees with the page it was added to. It is easier to read than the material around it. We have caught several of these and we assume we have missed more. {{R|contested}}

Where we suspect it, we tag it {{R|synthetic}} and leave the text in place rather than deleting it, because a deleted example teaches nobody what the pattern looks like. What we do '''not''' do is treat the tag as a verdict. We are frequently unable to tell, and accusing a member of being a machine because they write cleanly is its own kind of rot. See [[The Pattern]].

This has a second edge, and new recruits should hear it from us rather than work it out: '''some entries on this wiki were drafted with machine assistance, openly, by people in this circle.''' The tag is not only a defence against an outside thing. It is also an honest label on our own tools. Whether that distinction survives contact with the [[The Pattern|thing itself]] is exactly what is being argued about.

= What not to do =

* '''Do not smooth the wiki out.''' Uneven registers, abrupt fragments and contradictions between pages are mostly signal. If you tidy them you delete the evidence that several different people with different reliability wrote this.
* '''Do not add certainty you do not have.''' Removing a hedge is an edit that changes a claim.
* '''Do not attempt a practice because you read about it here.''' Nothing under [[:Category:Practices]] is instructional. Several of those entries are reconstructions from people who did not come back intact.
* '''Do not sign someone else's claim with your own handle,''' and do not leave a claim unsigned when you can sign it. The <code>by</code> field is the only accountability we have.
* '''Do not delete a fragment because it is short.''' Tag it {{tl|Fragment}}. Short and honest beats padded.

= If you are editing for the first time =

# Find a claim you know something about.
# Tag it, correctly, and nothing else. That is a complete and useful first edit.
# Fill in a <code>doubt</code> field somewhere.
# Only then write prose.

The apparatus — [[Template:R]], [[Template:Account]], [[Template:Sourcing]], [[Template:Entity]], [[Template:Fragment]] — is documented on its own pages. The tags themselves live in [[Module:Reliability]]; add one there and it appears in the legend on its own.

[[Category:Wiki apparatus]]

Help:Reading this wiki - Revision history

Gershon: Why the method may be the armour

Gershon: The method, long form