michael_nielsen

Ultimately, though, you want money to be supporting the best new models (and successor models, when those are found). And so far we haven't found a really good decentralized way to support preprints and whatever better models come after preprints.

The arXiv budget last year was ~$2mil. The American Physical Society's journal budget was ~$40mil. In terms of the social benefit I think that's inverse from where it should be. So: what would a good market structure look like, one where arXiv naturally outcompeted the APS?

Note: I'm really reluctant to want a model where the arXiv is simply locked in by central fiat. Better models than arXiv are certainly possible, and it'd be nice to encourage the emergence of those as well.

Excuse me: the APS's journal _revenue_ (not budget - expenses were about 30 mill, so their profit was about 5 times the arXiv's entire budget).

Yes, fully agree – I was already thinking of adding that PDF preprints are an end-run that is only a stopgap until better models for eliciting investment in infrastructure are made so that researchers can write & publish machine-readable interactive data-linked etc. documents.

There's plenty of money around, e.g. wasted on >20% Elsevier and Wiley profit margins based on subscriptions, the question is how to redirect it to publishing infrastructure investment that avoids free-rider problem resulting in not enough money being spent.

(Not to say that all profit is waste; important in capitalism to incentivize investment – rather, it's just that these very high levels of profits are based on a non-functioning market, so companies don't have much incentive to improve publishing platforms).

We should probably talk more about the market, because you're taking as proven that the market is non-functional & deriving conclusions based on that, but I don't think it holds entirely.