Today I’ve realized that there exists an attack vector on my valuation system of scientific publications and software packages.
Somebody may publish a low or mediocre quality scientific article and then build an exponentially growing business around it: He could write a growing number of articles directly and indirectly citing it. The articles could be low-quality to decrease the cost of their authoring but quite scientific in method (e.g. they would be things like calculation of the average weight of a key for a doorlock in Brazil). The cost of authoring such meaningless scientific articles in modern world is rather low. So, the intruder of my system could get a big (exponentially!) growing number of citations and so high rating of the first article in the chain. Because his growth is exponential, he is likely to overgame even best scientists and “steal” the money.
Some bad ideas of mitigation that come to mind:
- Give priority to articles published in reputable journals (it does not work because the prestige of a journal is again measured by citations, and the likewise trouble appears again).
- Use AI (to train it we already need to have a measure method).
The idea that I hypothesize to be an effective method of of mitigation:
Publication should be paid and high-cost. (It does not necessarily mean that an author pays for publication, a publisher could do.)
That is, the evaluator needs to give higher rating to paid publication. (We would lock such paid money in a specific contract(s).) As a a positive side effect, this strategy also makes possible to identify who is the publisher.
Then investing into a high number of low quality articles could become a non-competitive strategy compared to writing a good article that could be cited by independent parties.
Attacker’s counter-measure: Again produce an exponential growth of a low-quality article paying much for each its derivatives.
- Some businesses publish not to get direct return for the publication but for side effects (such as their commerce site’s promotion). So many business will invest (much) money into publishing competing with the above mentioned Ponzi scheme. Also nonprofit/generous purposes.
- Some people just want to get famous and will invest their personal money.
So, I hypothesize that the effect of low cost of writing low-quality article will be outweighed by the effect of the number of businesses citing high quality articles being big.
A side effect of the above strategy is that we get a link with real economy, that I consider good. (Well, does this exclude such works as constructions of time-machines and warp drives or planetary radioactivity cleaners that could not be implemented after 100 years?)
The above seems to be very bad for an author, because even if he/she himself does not pay, high cost necessarily makes publishers very selective. But consider that a publisher may initially invest a small amount then it or other publishers add more. Also the author may invest a small amount himself initially. That’s much better than modern $10000 entry to publish a monograph.
The above strategy also has other positive effects:
- No need to artificially filter out self-references and affiliated references: If an author pays for a self-reference, it is a worth business despite of being directed onto himself.
- The need for checking scientificity possibly (partially) disappears. No need to directly use people’s biased opinions to check scientificity of an article. So this strategy can be applied also to journalism in addition to scientific publications.
Obviously, for the above to work, the return from investing into a publication should be negative, if it is not cited by others.
So, the algorithm of valuing an article becomes very simple (not need to be Google):
- When paying C for a publication, transfer A*C where A is a coefficient to the cited article (how to spread, when several articles are cited). A < 1, A could depend on “how much” and “how significanty” it’s cited, but that’s subjective. After this transfer, recursively transfer to the indirect cited sources (to ensure the money does “multiply” indefinitely when the length of the chain of sources is big).
- Mint the proportional numbers of a conditional token to the publisher and to the author. (How to determine author reliably? We could use cryptographic signatures and not pay pubishers not conforming to such a cryptographic protocol.)
So, publishers can do business splitting investments between above mentioned publication charges and SEO. This makes investment into SEO (not only of science but of journalism, too) less profitable and therefore SEO less predatory.
We could have several duplicated schemes of splitting A between sources:
- by author’s decision
- equally (to reduce the incentive of authors to discriminate)
- by how many times the cite is meet in the text
- by an AI