According to Keltham's teachers on the subject of how to tackle Very Serious Problems, he should first try to collate all the points or observations that are potentially important:

- Observations or apparent facts that seem confusing even now;
- Things that stuck out at the time;
- Everything that caused him to update a lot because any of that could be an adversarial Conspiracy move;
- Anything interesting that was said early on; before the Conspiracy had time to plan a lot, or they knew what kind of target Keltham represented, or knew what was important to optimize-over about him.

Then Keltham should organize that first disorganized list into rough categories, using a computer that he does not, in fact, currently possess.

Then Keltham needs to stare at that list, look for intersections and correlations, write down a lot of possible questions or confusions.

Then Keltham is supposed to organize that rough list of confusions and questions and look for intersections between those.

Then Keltham is supposed to try to figure out plausible Conspiracies such that a lot of those questions would be resolved, and make note of the ones that are still unresolved for each Conspiracy theory.  He's also supposed to figure out plausible Ordinaries that would resolve those questions, if there's more than one notion of what the Ordinary world would have to look like, though in the Ordinary world he can just ask.

Then he's supposed to assign priors and sweep through all of his observations for all of his remaining plausible theories - once those theories have been pre-updated into latent detail from the evidence (and appropriately penalized in priors of those latent details), so that the evidence becomes independent over each observation; rather than observations updating the particular details of Conspiracy or Ordinary in a way that changes future likelihoods, in which case the pieces of evidence are no longer independent.

This sweep through all the evidence needs to include an estimate of the cumulative gradual evidence from all the minor occasions where something Conspiracy-ish could have happened but didn't, all the things that aren't confusing but could have been.  Because if you only look at the confusions given Ordinary, as will all tend to weigh against Ordinary, that's a methodology that will always conclude in favor of Conspiracy.

And then Keltham is supposed to throw out the final numbers and go with his brain's intuitive sense, once he's actually forced his brain to weigh all the pieces of evidence.

According to Keltham's teachers on the subject of how to tackle Very Serious Problems, he should first try to collate all the points or observations that are potentially important:

- Observations or apparent facts that seem confusing even now;
- Things that stuck out at the time;
- Everything that caused him to update a lot, because any of that could be an adversarial Conspiracy move;
- Anything interesting that was said early on, before the Conspiracy had time to plan a lot, or they knew what kind of target Keltham represented, or knew what was important to optimize-over about him.

Then Keltham should organize that first disorganized list into rough categories, using computer-based outlining software that he does not, in fact, currently possess.

Then Keltham needs to stare at that list, look for intersections and correlations, and write down a lot of possible questions or confusions.

Then Keltham is supposed to organize that rough list of confusions and questions, and look for interactions between those.

Then Keltham is supposed to try to figure out plausible Conspiracies such that a lot of those questions would be resolved, and make note of the ones that are still unresolved for each Conspiracy theory.  He's also supposed to figure out plausible Ordinaries that would resolve those questions, if there's more than one notion of what the Ordinary world would have to look like; though, in the Ordinary world, he can just ask.

Then he's supposed to pre-update each of those theories into 'latently detailed' background stories that would, ideally, make all the particular observations have likelihoods with no further dependencies between each other - rather than observations needing to update the particular details of vague Conspiracy or Ordinary pre-theories, such that later observations would have likelihoods entangled with earlier observations.  Of course, this pre-detailing step must further penalize the priors for each added detail.

Then he's sweep through all of his observations for the pre-detailed versions of all of his remaining plausible theories.

This sweep through all the evidence needs to include an estimate of the cumulative gradual evidence from all the minor occasions where something Conspiracy-ish could have happened but didn't, all the things that aren't confusing but could have been.  Because if you only look at the confusions given Ordinary, as will all tend to weigh against Ordinary, that's a methodology that will always conclude in favor of Conspiracy.

And then Keltham is supposed to throw out the final numbers and go with his brain's intuitive sense, once he's actually forced his brain to weigh all the pieces of evidence.

According to Keltham's teachers on the subject of how to tackle Very Serious Problems, he should first try to collate all the points or observations that are potentially important:

- Observations or apparent facts that seem confusing even now;
- Things that stuck out at the time;
- Everything that caused him to update a lot, because any of that could be an adversarial Conspiracy move;
- Anything interesting that was said early on, before the Conspiracy had time to plan a lot, or they knew what kind of target Keltham represented, or knew what was important to optimize-over about him.

Then Keltham should organize that first disorganized list into rough categories, using computer-based outlining software that he does not, in fact, currently possess.

Then Keltham needs to stare at that list, look for intersections and correlations, and write down a lot of possible questions or confusions.

Then Keltham is supposed to organize that rough list of confusions and questions, and look for interactions between those.

Then Keltham is supposed to try to figure out plausible Conspiracies such that a lot of those questions would be resolved, and make note of the ones that are still unresolved for each Conspiracy theory.  He's also supposed to figure out plausible Ordinaries that would resolve those questions, if there's more than one notion of what the Ordinary world would have to look like; though, in the Ordinary world, he can just ask.

Then he's supposed to pre-update each of those theories into 'latently detailed' background stories that would, ideally, make all the particular observations have likelihoods with no further dependencies between each other - rather than observations needing to update the particular details of vague Conspiracy or Ordinary pre-theories, such that later observations would have likelihoods entangled with earlier observations.  Of course, this pre-detailing step must further penalize the priors for each added detail.

Then he's supposed to sweep through all of his observations for the latently-detailed versions of all of his remaining plausible theories.

This sweep through all the evidence needs to include an estimate of the cumulative gradual evidence from all the minor occasions where something Conspiracy-ish could have happened but didn't, all the things that aren't confusing but could have been.  Because if you only look at the confusions given Ordinary, as will all tend to weigh against Ordinary, that's a methodology that will always conclude in favor of Conspiracy.

And then Keltham is supposed to throw out the final numbers and go with his brain's intuitive sense, once he's actually forced his brain to weigh all the pieces of evidence.