# The way to get around journals is to focus on non-paper artifacts There is no first principles reason why the Scientific Paper in its modern form (a ~5-30 page written document primarily made of words with a fairly specific structure) should be a unit of knowledge or research output. Of course, there are many good things about it, it has had hundreds of years to evolve, and it is a useful [[Schelling Point]] so we should take [[Chesterton’s Fence]] very seriously here. Perhaps it’s also useful to distinguish between a paper as a written document that is meant to convey knowledge, and a Scientific Paper as the specific artifact of credit, knowledge transfer, and stake seeped in history and institutional implications. Scientific Papers have become tightly coupled to an entire complex of Scientific Journals, Grants, Academic Careers/Tenure, and Peer Review both of the papers as a worthy artifact and of grant applications. This tightly coupled system creates a [[Wicked Problem]] because any piece of the system exerts a restoring force against changing any single piece of the system. If you try to change Scientific Journals (say by getting people to publish Scientific Papers online or in a new journal) the rest of the system will punish you by not giving those papers credit which damages careers and reduces the chance of getting grants. The coupling means that you can’t change any piece without shifting the system to a new, (hopefully better) equilibrium. How do you do that? Another framing might be to look at what the system’s implicit purpose instead of explicit purpose is. [[Complex emergent systems have purposes not causes and effects]]. Instead of the system’s purpose being to generate as much human knowledge and capability as possible, the system’s real/emergent purpose is to generate legible artifacts of work, award credit for them, and direct funding towards generating more legible artifacts. ([[The purpose of the Scientific Paper system is to generate legible artifacts of work, award credit for them, and direct funding towards generating more legible artifacts]]) Through this lens, the abstract goal is to shift the system’s emergent purpose towards its explicit purpose. Classic [[Clayton Christensen]] disruption theory tells us that innovations that aren’t just improving a system along its existing performance axes almost never win by going head on at the incumbent. Incumbents here being organizations, paradigms, and perhaps more importantly, applications. Minimills made steel for completely different applications that could usefully use recycled scrap steel. Similarly transistors were used in situations where vacuum tubes just didn’t work, etc. etc. Given this framing, I would argue that it’s important to focus on activities/applications that are currently poorly served by, or outside of, the system. The theory of change is that instead of standing up an entire new system, or trying to incrementally change the existing one, you peel off the poorly served bits of the existing system, build a better way to serve them and then pull the entire system in that better direction because people can now see a different, better, local maximum. Non-paper artifacts may be just the underserved application to focus on. Until at least the latter half of the 20th century, word-based papers (and books) were by far the best possible artifact. Paper was the cheapest and easiest way to transmit information at scale, figures and images were hard to both create and print (remember the photocopier wasn’t commercialized until after a human had already been in space). If they weren’t printed, large data sets had to be stored on punch cards, video on reels, etc. We no longer live in that world. If the digital revolution did one thing it made creating and disseminating different kinds of media (as long as it can be converted in bits) cheap and easy. As a result research can create many useful artifacts: datasets, code, podcasts, explainers, videos, audio, simulations, tools, CAD models that can be 3D printed, the list goes on. Not all of these are digital nor is that research *didn’t* produce them before, but modern computing has absolutely unlocked the ability to communicate about them without a paper. [[We need more scientific artifacts than papers]]. And yet, under the current system none of these other artifacts ‘count’ unless they are coupled to a Scientific Paper. Put aside for a moment the fact that only having Scientific Papers count forces non-paper artifacts to re-enforce the current system. A (more?) insidious problem is that the need for artifacts to couple to Scientific Papers puts constraints on those artifacts and the incentives to create them. For example, one criterion of Scientific Papers is that they “contribute novel insights to the literature.” Datasets can be painstaking and expensive to collect. They can be incredibly valuable for creating knowledge and capability but in and of themselves, they do not contribute a novel insight. So in order for a dataset to couple to a Scientific Paper, its creators need to both create the dataset *and* use it to generate a novel insight. The researchers will then get credit and attention primarily[^1] based on the Scientific Paper around that insight, rather than the dataset itself. So if the dataset is great but the insight isn’t particularly profound, the entire work bundle might be published in a lower-status journal and reach fewer people. Needing to have a Scientific Paper front for a dataset means that the bulk of the contribution is basically a buried lede. Additionally, the vast majority of the search and discovery infrastructure for scientific knowledge is built around Scientific Papers so disseminating the dataset is just harder. The example focused on datasets but the same dynamics apply to other non-Scientific Paper artifacts. I listen to several podcasts by professors and in them they explicitly lament that they get almost no career credit for doing it and as a result the podcast is a pure passion project. The one bright spot is that the current system’s terribleness at serving non-Scientific Paper artifacts provides an opportunity to do some [[Clayton Christensen]] style disruption.[^2] Non-Scientific Paper artifacts are all the things that can be built with recycled scrap steel and a system to support them are the [mini-mills](https://www.nytimes.com/1981/09/23/business/the-rise-of-mini-steel-mills.html). A natural follow-up question is “what artifacts should this side-system focus on?” Aggressively, I would argue the answer is ‘none.’ [[A system of credit and funding for non-Scientific Paper artifacts should be artifact-type agnostic]]. ### Related * [[Scirate]] * [Swipe right for science: Papr app is ‘Tinder for preprints’ : Nature News & Comment](https://www.nature.com/news/swipe-right-for-science-papr-app-is-tinder-for-preprints-1.22163) * [[Most scientific papers are hard to read]] * [[‘I published a paper on that’]] * [[Things that are not paper-worthy enough for academia and not product-focused enough for a startup]] * [[Labs implicitly judge research through the sense of dollars per paper]] * [[Another downside to importing graduate students to work in labs is that it makes papers less intelligible]] * [[The peer review and citation system incentivizes people to work on things that other people think is interesting]] * [[It is much easier to share an artifact than a process]] [^1]: Obviously there are notable exceptions here. [^2]: I cringe a little at using the term “disruption” because of its overuse but I think in this case I think it’s actually just the right textbook term. [Web URL for this note](http://notes.benjaminreinhardt.com/The+way+to+get+around+journals+is+to+focus+on+non-paper+artifacts) [Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/The+way+to+get+around+journals+is+to+focus+on+non-paper+artifacts)