I enjoyed Jon Atack's talk at PlanB in January (#1447597) but felt that it was a little abbreviated -- conference talks having their necessary time restrictions.
To remedy this Bob Burnett, CEO of Barefoot Mining (the only miner to have mined a block signalling for BIP-110 - #1445509) interviewed Atack on his show, Old Man Yells.
Here are a few notes from listening to the episode:
First of all, there is a lot of introductory and background discussion in the first half hour or so, where we hear about Burnett and some of Atack's background. It's interesting to hear about their careers as they are both quite accomplished, but it may be the case that you could skip the first half hour.
A lot of the discussion elaborates on the points Atack made in his PlanB talk: how funding works for developers working on Bitcoin Core, and some of the social or political dynamics that Atack's sees as playing a role in how development on Core is done.
Another interesting thing Atack points out is that there is grant money available for developers who are interested in working on alternative implementations to Core, but mostly there has been a lack of qualified applications. OpenSats is currently funding one developer who is working on libbitcoin, but it seems that they would be willing to fund others to work on Knots.
The most interesting part of the conversation to me was about the world where we have multiple bitcoin implementations. Burnett (formerly the CTO of Gateway) makes the analogy to the personal computing industry circa the early 2000s. He suggests that personal computing was similar to Bitcoin, with each PC company (Gateway, Dell, Toshiba, etc) needing to maintain a loose consensus around PC designs so that things like USB or other standards could be adopted and utilized across the industry.
Burnett talks about how the various PC companies were required to pay attention to their stakeholders and respond to their needs because otherwise customers would leave to a competitor. Burnett then likens Core to Microsoft Windows, who Burnett says operated with much less attention to customers because they had so much market share.
At one point Burnett says:
When I talk to other miners...i have a lot of friends obviously who are in this space...who are not the pub-co's by the way, not the big miners -- the small miners and I do not know of a single one who has spoken to someone from Core or been asked their opinion about whether or not what's being worked on is something they agree with.
Both Burnett and Atack have many good things to say about Core as well, but the context of the conversation is one which is attempting to address some problems they both see in Bitcoin Core's role in Bitcoin as a whole.
It's a bit of a long listen, but if you have the time it's an interesting perspective that tends perhaps to get hidden by the drama and tribalism of the Knots vs Core debate.
Well, Bob as had conversations with a few developers, and there is a discussion between him and Pieter on the mailing list, but not sure what exactly counts here.
That said this has been been a problem since way back, and something that I would genuinely like to remedy. We should be useful to miners, big and small alike. If someone happens to read this that is running a small operation, I'd like to hear what we could improve for them.
Thanks for some context there. I imagine one of the difficulties of being on the Core side of this is that everybody can see Core while not everybody can see all the many mining operations out there. What may have felt to Burnett like being ignored or disinterest, could actually just be too much inbound.
This could easily be a full time job for someone. Just going around, gathering and maintaining new and existing requirements, and advocating for them competently.
Certainly. and the same could be said for gathering feedback from node runners. During the interview they both commend @schmidty and BItcoin Optech for their work on this front.
Interesting background. Is Gateway still around? I remember considering buying a computer from them once, but never ended up doing it. (Forgot who I went with ultimately).
There's always gonna need to be a social consensus of some kind. No way around it.
That's fascinating. Did they say if it was because everyone wants to work on Core, or because there's a general lack of developer interest in Bitcoin at all?
I'm kinda afraid the answer is the latter. AI is probably sucking all the air out of the room for anything computer related
From my perspective it is the latter. There have been some significant recruitment efforts this year again, and so far not much has materialised from them. The project is not in a terrible place manpower wise either, it managed to attract some great talent the past few years. It is clearly getting harder to get new graduates to be interested in it, and with the past decade it has spent on the market already, many established developers that would be interested in it, are already captured in its orbit, or moved on to other things.
Check out 1:56:00 or so
Perhaps it's good if Bitcoin isn't the "hotness".
I wouldn't want the developers of Core to be interested in it because it's a good career move or looks good on a resume.
Yes, that probably means fewer applicants, and less qualified applicants. But for something that has a certain ideology attached to it, I do prefer devs that are somewhat more ideologically motivated.
Hot damn, I love me a moderate take.
Improving bitcoin's social processes is very fun to think about. IMHO money, infrastructure, and dev centralization are downstream of attention/influence centralization; attention is how money etc find their home. So how do we decentralize that or at least level that playing field?
While I'm sympathetic to the theory that multiple clients fixes this, I think winner-takes-all via network effects will continue to be the norm so long as humans are interdependent. Perhaps it's another case where credible exit, a competitive threat, is the best we can do.
I had a list of potential things to do when schmidty's v30 PR wouldn't get merged. My favorite one was: Host a repository of tested, vetted source patches and automate the regressions. Add a single dedicated EPYC box with some massive disks and you can run guix builds all day every day and provide proper binaries too.