Dear Jamie,
As agreed on the phone this letter will be the last for a while because we’ve agreed to start a new letter series on a friend’s new platform. You can’t respond to this letter any time soon so I don’t want to spend it arguing counterpoints, that would be unsportsmanly. Instead I’ll propose my ideal rules for entertaining new scientific theories.
Truly novel knowledge is, by definition, utterly unpredictable. Let’s imagine bucket A contains things we know. Then imagine bucket B contains things we don’t know explicitly, but we know how to eventually know it. Then bucket C has everything we don’t know and don’t know how to know it. This is the unpredictable bucket.
I think most scientists think there is no bucket C because the scientific method is a technique through which we can eventually know anything (not necessarily everything). In contrast it is my opinion that there is no bucket B. Everything in bucket B should be put in A. What we know we know counts as existing knowledge. It’s inevitable B things will end up in A there just hasn’t been enough time, or we’ve been lazy or something.
So the most important distinction is A/C: what is and isn’t known. Probability can help within A and B, but cannot help us in how to search C.
It’s almost universal in science to think bucket B is the biggest and most important so the prevailing attitude is to focus on it. So when attempting to make progress the approach is one of careful quarantine and deliberate gradual iteration. That strategy sees the community desiring theories that are likely to help us sort out bucket B. This appears as a strong quarantine which I believe gets misidentified as skepticism: Do not allow theories that aren’t consistent with current knowledge (only things from A and B). Once a theory gets through these stringent customs it has essentially done its work. Scientists are generally free to explore and publish even the most humdrum ideas to peers with varying standards of rigour.
But if, like me, you think A and B are one already-known bucket that only allow provincial progress, this quarantine does nothing to help the crticial A/C distinction. Popper’s approach to customs swaps the vigilance around. Loosen the quarantine to you have a hope of finding things in C. Then make up for this laxity with the strictest possible truth tests in the day to day running of the academy.
The strictiest possible test is based on the strictest interpretations of truth within logic: What is the most reliable test for truth? There are debates over whether something like induction can be used to extrapolate observations. The most charitable position on this is “jury’s out”, but if we’re in the accuracy game we can’t use something whos truth is disputed to detect truth. Bin it! Some might calculate probabilities to determine likelihood of truth of a theory. Who knows if this is valid, but it’s at best shaky ground. Into the bin!
Continuing to clean house in this manner, disposing of even the least hopeful parts of logic, leaves you only with deduction. The idea that, given statements or facts or theories, that they only imply what they say where they say it. That’s it. Deduction has the added side benefit that only the boldest claims can be judged, universal claims like “for ALL x, y happens”. This is the kind of knowledge we want. Knowledge with unpredictable exceptions isn’t knowledge.
Universal claims are unexhaustible so we can’t test every case for truth. But can be disproved with a single exception.
For all x, y happens.
[perform test x_1 expected to give y_1]
result: x_1 → NOT(y_1)
Thus NOT (For all x, y)
This is the strictest test on the boldest theories. It is the only necessary condition for holding theories accountable to reality that we have, so let’s use it as our standard for rigour. Armed with that we can open the borders!
Scientists much prefer the quarantine at the door option. The strict tests of the alternative make the criterion of progress too hard to deal with, while quarantine has the lovely side benefit of keeping out the riffraff. But such strict border protection is a symptom of fear mascquerading as a legalistic skepticism. If outsiders play by our strictest rules, what’s there to be afraid of?
Till later,
Mat