michael_nielsen

Brokenness 2: The big deal. Journals used to be sold a la carte. But beginning in 1996 was the era of the era of "the big deal" (e.g. https://poynder.blogspot.com/2012/10/open-access-in-uk-reinventing-big-deal.html … ). Instead of buying journals individually, libraries bought them in big bundles, sometimes of thousands of journals.

In many ways this was good. The internet meant that publishing infrastructure could be centralized, giving big publishers economies of scale. Bundling passed some of those economies of scale onto customers.

But it had many negative consequences. It meant that journals became a commodity bought in bulk, typically distinguished only by a very imperfect brand / quality marker like impact factor.

This was happening at a time when experimental new ideas should have been flourishing. At such a time you don't want things bought in bulk, you want them bought bespoke, based on highly idiosyncratic and individual criteria.

And so instead of competing on the basis of amazing new product types, increased product quality, & increased access, publishers instead compete by achieving economies of scale, driving down operating costs while maintaining revenue, improved sales, & maximizing brand lock-in.

This is reflected in many ways: most obviously, the many mergers and acquisitions of publishers, giving increased economies of scale. This has the very unfortunate by-product of reducing competition.

It also means that many (not all) of the people running scientific publishing are business people who specialize in managing operations (driving down operating costs while maintaining revenue), and in sales and marketing

An example: IIRC Derk Haank, the CEO of Elsevier from 1998-2004 and of Springer, later Springer-Nature, from 2004 to 2017, did his PhD on economies of scale.

Nothing intrinsically wrong with this. But we're at a time in history where the socially beneficial act isn't driving down operating costs while maintaining revenue. It's producing marvellous new tools, increasing access, etc. Current market structure isn't supporting this well

Brokenness 3: The lack of growth models for the best new ideas. An example is the arXiv preprint server. It's one of humanity's great achievements of the past 30 years. Just in economic terms, over the long run it will generate trillions of dollars in social utility for humanity

If it captured just a tiny fraction, the arXiv would have a budget of tens or hundreds of millions of dollars a year. Instead, the arXiv has struggled to make budget for much of its existence. It can't grow or innovate the way it should, & changes at a glacial pace.

No criticism of the arXiv intended - this is a consequence of a systemic factor: the lack of good growth models that enable great services to grow and change and improve.

You see this pattern repeated over and over for a tonne of new tools. Great new tool, no growth model. And so they stagnate and languish.

One response is to say "Oh, the NSF [or whoever] should give a lot more funding."

I'm sympathetic, but only as a stopgap. It's not a good long-run solution. If centralized authorities are providing money, you end with the arXiv (or whoever) as a de facto incumbent, being funded by decisions made by a small group of ppl. This is a recipe for stagnation, at best

What you really want is to encourage the arXiv to grow & innovate, _and_ also to fund potential competitors who aim to do even better than the arXiv. And, if things are healthy, they will replace the arXiv.

So, to come back to where we started: are for-profits bad? Should we aim for a not-for-profit future in scientific publishing?

I hope it's clear these questions miss the point. Better questions are: what's the growth model for innovation? Is the market set up to enable the flourishing of many good new ideas that will benefit humanity? At the moment, it's not doing a great job, in my opinion.

Instead, incumbent organizations maximize revenue in ways that do serve some social job (journals are good things), but far less than could be done, and often with a lot of negative behaviours. This is true both of for-profits like Elsevier, & of many not-for-profit publishers

Go take a look at the American Chemical Society, a not-for-profit publisher with billions in revenue. Historically they've been far more hostile to ideas like open access and open data than Elsevier & the other large for-profit publishers. https://en.wikipedia.org/wiki/American_Chemical_Society#Controversies …

Many other not-for-profit society publishers aren't much better. Any serious argument that "for-profits are bad" needs to engage with this fact.

So, what to do? A lot of progress so far has come from things like the Bermuda Principles, whereby the NIH and Wellcome Trust essentially forced biologists to share human genome data. https://en.wikipedia.org/wiki/Bermuda_Principles …

Or the NIH public access policy, which requires NIH-funded research papers to be shared after an embargo period: https://publicaccess.nih.gov/ 

I'm a huge fan of these and similar actions, and of people and organizations like @hjoseph, @petersuber, @SPARC_NA and the many, many others who helped them become a reality. These are some of the most important accomplishments of humanity in the past decades.

At the same time, over the long run we want to avoid running things through centralized control. Command-and-control economies have a terrible historical record, and usually end up inhibiting innovation.

That's my view of the problems. What of the solutions? How to create a healthy competitive marketplace in scientific publishing?

There's not going to be a silver bullet. It's going to require hundreds of changes. One crucial change is getting existing funders to take software tools seriously. Budgets for tools, for programmers, for long-term maintenance, & VC / grants for new organizations to develop tools

And payment for services needs to align incentives: the people benefiting from the services should be paying for them, to set up the virtuous feedback loop: genuinely better service => more revenue.

This is a tough problem for open * solutions (open access, data, code, collaboration). It still hasn't been solved in the world of open source software. Though companies like Kickstarter and Patreon and ideas like dominant assurance contracts are making progress in this space.

Still, I'm optimistic we can solve these problems. Danny Hillis has observed that "there are problems that are impossible if you think about them in two-year terms - which everyone does - but they're easy if you think in fifty-year terms." I think this is a problem of this type!