# [Question] What makes a scientific fact ‘ripe for discovery’?

The ex­is­tence of mul­ti­ple dis­cov­ery seems to sug­gest that there are cer­tain fac­tors that make sci­en­tific facts ready to be dis­cov­ered. What are these, fac­tors, and how could one mea­sure them?

• Mul­ti­ple an­gles of attack

Let me warn you, “im­por­tant prob­lem” must be phrased care­fully. The three out­stand­ing prob­lems in physics, in a cer­tain sense, were never worked on while I was at Bell Labs. By im­por­tant I mean guaran­teed a No­bel Prize and any sum of money you want to men­tion. We didn’t work on (1) time travel, (2) tele­por­ta­tion, and (3) anti­grav­ity. They are not im­por­tant prob­lems be­cause we do not have an at­tack. It’s not the con­se­quence that makes a prob­lem im­por­tant, it is that you have a rea­son­able at­tack.

One rea­son­able at­tack makes the prob­lem ap­proach­able. If there are mul­ti­ple rea­son­able at­tacks, at least one suc­ceed­ing be­comes more likely and fur­ther they can ex­change in­for­ma­tion about the prob­lem mak­ing each at­tempt more likely on its own. If we switch to con­sid­er­ing thor­oughly un­der­stood prob­lems, we usu­ally have mul­ti­ple good solu­tions for them (like mul­ti­ple proofs in math­e­mat­ics, or de­tec­tion from differ­ent kinds of ex­per­i­men­tal ap­para­tus in sci­ence).

So if I am go­ing to rank open prob­lems by the like­li­hood they will be solved, my prior is a list or­dered by the num­ber of ways we know of to at­tack each prob­lem. Without any other in­for­ma­tion, a prob­lem with two rea­son­able at­tacks is twice as likely to be solved as a prob­lem with only one.

Then we could con­sider up­dat­ing the weights of differ­ent kinds of at­tack. For ex­am­ple, if one re­quires very ex­pen­sive equip­ment, or very rare ex­per­tise, I might ad­just it down. On the other hand, if there are two differ­ent at­tacks but the re­la­tion­ship be­tween those two ap­proaches is oth­er­wise very well un­der­stood, then we might not treat them as in­de­pen­dent any­more and fac­tor in the ease of shar­ing in­for­ma­tion be­tween them but also that they will prob­a­bly suc­ceed or fail to­gether.

We can also con­sider the prob­lem it­self, but I feel like look­ing at the refer­ence classes for a prob­lem largely boils down to a way to search for rea­son­able at­tacks, where any at­tack which worked for a prob­lem in the refer­ence class is con­sid­ered a can­di­date for the prob­lem at hand. But as I think of it, I’m not sure it is com­mon to do a sys­tem­atic eval­u­a­tion in this way, so high­light­ing it as a spe­cific method for find­ing at­tacks seems worth­while.

• Ini­tial Brain­dump (hope­fully will edit)

Knowl­edge de­pen­den­cies (alge­bra be­fore calcu­lus)

Ne­c­es­sary tools? (Did peo­ple who made si­mul­ta­neous dis­cov­er­ies use the same tools?)

The re­search com­mu­nity for that do­main? (How much com­mu­ni­ca­tion is there? How dense are the con­nec­tions be­tween peo­ple?)

How new the field is. Whether there was a sud­den jump in the num­ber of re­searchers.

How fre­quently dis­cov­er­ies hap­pen in the field.

Whether a ma­jor dis­aster or other event is ob­struct­ing sci­en­tific progress from be­ing made at the time.

Whether the ex­is­tence of si­mul­ta­neous dis­cov­er­ies is just an ar­ti­fact of cher­ryp­ick­ing biases