Re:Number theorist fears many proofs widely considered to be...
by bob1029 on Sunday September 29, 2019 @01:31PM
(#21108344)
I feel the same way when I look at many of these papers.
I get that I don't have a post doctorate in super math or whatever the bleeding edge academic ivory tower game is these days, but at the same time I feel that if 90% of the research cannot convey any practical description of real-world implications, or at least how it relates to other abstract things that DO have real-world implications, maybe there is some BS house-of-cards fantasy game going on.
From what I understand, there are some financial benefits to cranking out research papers. Perhaps researchers are incentivized to build layer upon layer of abstraction so they can spin any tale they desire in order to turn a quick buck? What better field than math? The purest and most non-tangible of fields.
Another perspective I have on this - Virtually all computer technology or software papers I read are typically understandable to some degree, despite my not having the highest credentials in my field. It would seem the constraint of "if this is in a compsci paper, it needs some reference implementation pseudocode" seems to help keep people honest. If the computer can run it, at least you can view things in a more concrete (albeit still somewhat abstract) manner. Especially imperative code examples. These quickly take very abstract algorithmic concepts and transform them into a step-by-step understanding. If the paper is total bunk, you can usually tell pretty quickly from the results on the computer.
Automated proofs, where feasible, seem like a very reasonable requirement for published math papers. My ignorance at the higher levels of this field fails to inform me if all math papers could be proven automatically, or if there is some more 'complex' realm of reasoning unsuitable for classical/quantum/etc verification. Perhaps this should be a good constraint regardless - If you can't implement your algorithm/proof on a computer, where are you headed with it anyways? Surely piling another 500 papers worth of layers on top isn't going to move you any closer to a practical outcome.
I get that I don't have a post doctorate in super math or whatever the bleeding edge academic ivory tower game is these days, but at the same time I feel that if 90% of the research cannot convey any practical description of real-world implications, or at least how it relates to other abstract things that DO have real-world implications, maybe there is some BS house-of-cards fantasy game going on.
From what I understand, there are some financial benefits to cranking out research papers. Perhaps researchers are incentivized to build layer upon layer of abstraction so they can spin any tale they desire in order to turn a quick buck? What better field than math? The purest and most non-tangible of fields.
Another perspective I have on this - Virtually all computer technology or software papers I read are typically understandable to some degree, despite my not having the highest credentials in my field. It would seem the constraint of "if this is in a compsci paper, it needs some reference implementation pseudocode" seems to help keep people honest. If the computer can run it, at least you can view things in a more concrete (albeit still somewhat abstract) manner. Especially imperative code examples. These quickly take very abstract algorithmic concepts and transform them into a step-by-step understanding. If the paper is total bunk, you can usually tell pretty quickly from the results on the computer.
Automated proofs, where feasible, seem like a very reasonable requirement for published math papers. My ignorance at the higher levels of this field fails to inform me if all math papers could be proven automatically, or if there is some more 'complex' realm of reasoning unsuitable for classical/quantum/etc verification. Perhaps this should be a good constraint regardless - If you can't implement your algorithm/proof on a computer, where are you headed with it anyways? Surely piling another 500 papers worth of layers on top isn't going to move you any closer to a practical outcome.