Re: Academic Involvement in Hyperledger
No, not the architecture working group. While it is wonderful, It is not a research-based group.
I think this new “[*]group” should have two major focuses: (1) discussing research topics, ideas, results, and problems around Hyperledger and (2) incorporating these into Hyperledger projects. These two topics are officially out of scope of the architecture working group. We have had an occasional academic or two drop by the architecture working group, but they never stick around because it is not academically focused. If we just tell academics to join the architecture working group, we will never get any traction.
From: Middleton, Dan [mailto:dan.middleton@...]
Sent: Tuesday, July 9, 2019 10:20 AM
To: Montgomery, Hart <hmontgomery@...>; Arnaud Le Hors <lehors@...>
Subject: Re: [Hyperledger TSC] Academic Involvement in Hyperledger
Architecture working group?
From: <tsc@...> on behalf of "hmontgomery@..."
Thanks for the response. You are correct: I had, in fact, envisioned this as a bridge between maintainers/developers and researchers: researchers could contribute relevant research (and code) to maintainers for implementation in the project. Maintainers and developers, in turn, could contribute back interesting research problems to the researchers. This could serve as a mutually beneficial relationship where researchers get new ideas (and papers, of course) and maintainers get new improvements (and code). In addition, it would potentially get more academically inclined folks interested in contributing—lots of people like putting their ideas into practice. Currently we don’t have a lot of academic contributors. Lots of academics contribute directly to public blockchain stuff, and I don’t see why Hyperledger should be different.
However, I’m still not sure of what form this should take. It could be a working group—outputs could include code for projects, or at least RFCs/change suggestions (that the projects could implement or not)—or a SIG or something else entirely. What do people think?
My first reaction was similar to Chris's and Dan's - we don't need yet another group - but having thought about this a bit I'm actually in favor of creating a group dedicated
to academic people.
Thanks for all of the responses to my initial email. Rather than respond to all of them individually, I’ll try to address some points below. Hopefully this will be easier to follow than many different follow-ups.
From Virgil’s email:
When Ethereum had this same problem, we started the https://ethresear.chmessageboard to discuss problems and papers. Something similar might be useful to Hyperledger.” I agree!
From Brian’s email:
“We have long believed we need to engage academia, and have a formal process for doing so…” I spoke to people who tried to start this formal process and ended up somehow getting black holed (they filled out the paperwork/forms, but didn’t get back responses). This was a point of frustration and a reason why at least one person didn’t get involved further. That particular case should be rectified, though—thanks Marta!
“And true to form, we even have a mailing list.” Does anyone use this? I guess I should probably sign up….
“One challenge I've seen is that many academics aren't that familiar with open source in general - and aren't motivated to do more than public their papers and proof of concept code, since they are rewards for publishing, not for solving problems (at the risk of gross over-generalization).” This is certainly true for people that sit at their desk and do math all day. However, it’s far from the truth for good systems researchers. This point from Brian actually got brought up at the Dagstuhl seminar, but people didn’t agree. Mike Franklin (chair of the CS department at the U of Chicago) pointed out that he always liked to have his group contribute things back because it allowed them to find future research problems in a way that just writing the paper and going away did not. However, I will agree that most academics aren’t familiar with open source—other than Shahan Khatchadourian (of PegaSys), I think I was the only other person at this seminar with open-source experience, and most people were totally clueless. I think this means that we have an education issue with the blockchain research community.
From Chris’s email:
“What I would recommend is that we, as the TSC, take a hard look at contributor on-boarding of the various projects, suggest improvements to, or alignment of, the various projects processes and documentation and take a look at the various ways that a new user engages in Hyperledger generally to make sure that we are being as effective as possible in capturing new contributors.” I agree 100%. Even just posting the text from this email on the university page would be a good start!
From Vipin’s email:
“Please start a rocketchat channel called #research and publicise.” I’d support this as well.
From Dan’s email:
“Most if not all projects have contributors meetings listed on the HL calendar and CONTRIBUTING.md in the repos. Sawtooth’s for example points to: https://sawtooth.hyperledger.org/docs/core/releases/latest/community.html.” When people asked about contacting the maintainers or contributing, I mentioned this. No one was aware that these kind of files existed. There was also confusion related to the calendar—people weren’t sure which meetings they needed to attend. There are tons of meetings on the calendar these days! Some clarity on these would be great.
“Totally agree with Chris that another meeting group isn’t the best way to bridge the initial gap. In fact, I think siphoning that new-contributor interest into a meeting group is likely to be counterproductive.” I don’t agree with this. I think creating a group for researchers is likely to spur more researchers to attend and get involved rather than siphon away people. Ideally such a group would lower the activation energy for researchers to get involved and contribute and, in the long run, funnel them into the projects. I wouldn’t want to suggest a frequent or work-heavy meeting group—just discussions either biweekly or monthly. Can you explain your thought process more on this (i.e. tell me why I’m wrong ;))?
From Vipin’s next email:
“We have had several suggestions.” These are good ideas.
From Mark’s email:
“FWIW, the PSWG has three members* who were in academia and contributed to our first Metrics document. One of these folks also is a Caliper maintainer now.” There were several people at the seminar interested in Caliper. I suspect that it would be a big beneficiary of more academic involvement.
If you’ve made it this far, thanks for reading! I didn’t expect this to spur so much discussion, but I think it is something that, if we address properly, could really benefit Hyperledger. Thanks again for reading, and have a great day.