Or is it “CryptoTulips for Science”?
I’m a long time veteran of scientific computing, including the extremely tough problems of electronic publishing and knowledge dissemination, which requires good ways to deal with provenance and trust.
So I’m a bit surprised to learn that there is a magical solution to these insanely wicked problems: blockchain!
Specifically, the First International Conference on Blockchain for Science, Research and Knowledge Creation happened this month in Berlin.
This appears to be a conference of hammer-makers, so everything looks like nails to them. : – ) The “hammer” is the blockchain, and digital science looks like a box of nails.
According to the conference prospectus, the blockchain is more of a Swiss Army Knife, which can “to reestablish trust into scientific data”. Maybe it can “fix the reproducibility crisis”.
“As the ‘trust machine’ Blockchain in Science bears the potential to reestablish trust into scientific data. Some claim that it might even be good to fix the reproducibility crisis. New ways to rethink research subject privacy and whole data marketplaces are on the horizon. Blockchain might even play a large role in scientific publishing, where it questions the current role and business models of scientific publishers. New ways to incentivise peer-review or reproduction of results may arise. “
More plausibly, it might be useful for “data marketplaces” (assuming that scientists can afford to participate).
And it might also be a useful begging cup to help finance publishing and peer-review.
Much of the conference program is about blockchain technology (“get your cryptotuliips here!”), not so much about the alleged problems to be solved, let alone working solutions.
I’ll note that I don’t see any of the big names in Provenance or escience (e.g., Sensei Carole Goble really should have been a key note speaker, IMO) How can you talk about trust and reproducibility if you ignore the work that has already been done?
The thing is, blockchain qua blockchain offers little that can’t be done with conventional data systems plus public key cryptography. (I’ve said this before, and I’ll probably have to say it again.)
In fact, blockchain technology is a terrible fit for a lot of the problems, just as it is for other applications.
Science is a tiny, underfunded enterprise that does not need a global public blockchain. Science deals in weird, unique, often bulky data that ain’t never going to be on the actual blockchain. Reproducibility requires incredibly complicated records of information flows and processes, which could be recorded on a blockchain, but probably shouldn’t.
I’ll note that one of the most crucial operations of science is revision and retraction of errors. Blockchains cannot support that operation at all. Flooding the world with fake data that cannot be removed is not going to be a real advance.
IBM’s contribution to this field reflects this very fact: Zach Seward reports that IBM has registered a patent for using blockchain for “open science” . IBM’s idea of “open” science may not be the same as mine, but they certainly understand the problem. Their system is basically a change log for data, reports, and whatever. This approach tracks and makes a public record of who did what, including corrections and retractions.
The thing is, of course, that we already have change logs (see, perhaps Github, which is the fourth or more generation of this technology), and we have been using them for several decades now. We were working on these “digital notebooks for science” fifteen years and more ago . I assume that the new thing is implementing this on a blockchain which is technically clever but who knows if anyone actually needs it.
And, by the way, “science” is not a single enterprise, it is a bunch of small, inbred communities. I might have use for some kinds of data, but almost certainly have no way to even understand 99% of the data out there. So the blockchain will carry rafts and rafts of data that only a handful of people actually are able to use any part of.
It is interesting to consider that “scientific consensus” bears no resemblance to Nakamotoan “consensus”, because—it’s too complicated to go into here. The point is, scientific results are not valid of important because the author thinks so, or because of the number of downloads. The blockchain may assure accessibility and tamper resistance, but the evaluation of results will still work the old fashioned way.
For example, take a look at Wikipedia. It is a giant change log. There is a public record of who did what. This has worked amazingly well for a long time–without blockchain. Reimplementing it on blockchain would do nothing much, because the hard parts of Wikipedia are what the humans do.
I wonder if some of these notions about “incentvizing” publication, reviewing, and replication are a good idea or not. I understand why there is a temptation to scramble for funding, but it is a slippery slope to put science on a market driven model. This must inevitably distort what is done and published, rewarding trendy and politically favored topics, and starving less popular work. It may also be used to further cut public funding, on the excuse that “those scientists are raking in all that cryptocurrency”.
To me, this conference looks like a bunch of Tulip merchants trying to convince people to buy their magic CryptoTulips.
I may have to create a special CryptoTulip of the Year citation, for “type 3 CryptoTulips”. A “type 3” error is “asking the wrong question”, so a “type 3 CryptoTulip” is a confident solution based on misunderstanding of the actual problem.
- Zack Seward (2018) IBM Says Blockchain Can Power ‘Open Scientific Research’ in New Patent Filing. Coindesk, https://www.coindesk.com/ibm-says-blockchain-can-power-open-scientific-research-in-new-patent-filing/
- James Myers, Luigi Marini, Rob Kooper, Terry McLaren, Robert E. McGrath, Joe Futrelle, Peter Bajcsy, Andrew Collier, Yong Liu, and Shawn Hampton, A Digital Synthesis Framework for Virtual Observatories, in UK e-Science All Hands Meeting. 2008: Edinburgh.