I really don’t like “commit counters” because people who don’t know will inappropriately equate green boxes with productivity.
But it is useful for general signs of activity. So I think it’s stupid to say stuff like “foo has 120 commits and bar has 119; therefore foo is more productive”
But I think it’s a good starting point to think something like “Hrm, Bar only has 21 commits in a year and the rest of the team has 120. Let’s look into that.”
Sometimes there’s a good explanation (eg, “Foo makes a huge commit every two weeks, let’s talk to them about the benefit of daily check ins and branches”) but it also helps identify real problems.
I once worked with a company that was wrapping up a project. They have contractors 30 days notice to finish their work at the end of the release and go. One contractor went from regular activity to zero commits during the last month. Since I was supposed to integrate and code review his stuff, I asked if he would check in and he just said “yeah sure.” He left with no commits during the month. Not that there’s a specific number of commits that was good and he was obviously mentally checked out, but zero indicates something worth investigating.
It's interesting, because this tracks my take on code coverage metrics, speaking from a quality background.
There are only two intrinsically meaningful values: None, and less than yesterday. Former indicates no analysis, second indicates new code/branch/paths checked in without tests.
Past that, it's all relative. Major code coverage here and sparse there tells some kind of story, either about lopsided risk, lopsided difficulty, or lopsided process.
It’s funny because I like knowing it and find it useful but can almost never talk about it because it gets misinterpreted by pointy haired boss-types who want to track it, or practitioners who want to cargo cult it, or cynical people who think it’s worthwhile to game.
Office365 introduced some crazy usage metrics like number of minutes and “engagement” and whatnot. It was total garbage and not right (showing my first active as 1030, despite using mobile and sending tons of email and doc edits at 6am). But it had a metric about engagement through using at mentions in my direct report chain. Some manager thought this was good and praised people very engaged and instructed people to be more engaged. A few folks just at mentioned instead of putting names in their email’s To field. They shot to the top of this dipshit’s leaderboard.
"I'm just going to be explicit. In 2020, Henry created 12 issues, commented 25 times, and created 29 pull requests. This is across all Babel orgs."
"Sorry but that's DEFINITELY not $132k worth of work. Especially when there are other contributors who are working for free doing much more. Obviously everything isn't on GitHub, there's work in private but it is representative."
I understand that the tweeter is the original author of Babel - as such, I find this a little bit embarrassing.
The author originally had 100% control of governance and management decisions, and personally chose exactly how to delegate those.
A single person's salary (and the public disclosure of it) has created massive point of political tension for the project. I think that's a bigger problem than the perceived waste of $.
Yeah I also don't understand why this wasn't handled sooner. He said that in February the dev only made a small number of issues and no commits, I feel like the issue should have been caught there.
There's also a question of - what's the job description?
Henry makes it clear in a March 2018 blogpost that he doesn't see it as a primarily coding job, and his Twitter bio describes him as 'steward', not 'developer'.
OP called out PRs and comments, not just code. I’m not familiar with this project, but there’s a lot of stuff required for projects that isn’t coding but still shows up in GitHub’s activity. Even basic stuff like doing demos and updating customer stuff shows up as updates to the web site and notes on presentations and stuff.
Unless someone’s job is just sending email and talking on the phone, I’d expect some visibility. Not for tracking purposes, but for community engagement since it’s the best way to clarify requests, produce roadmap info, etc.
Whatever happened, this is not the way to air this information, or the right medium to judge someone (1 tweet).
Sebastian makes himself look bad here, and added to it that he is leaving babel, to start a competing product with VC funding, makes his public take-down of babel even worse.
Yes, but he also says he’s mentioned it privately many times. When the appropriate channels are exhausted, this seems appropriate for an OSS project where transparency is part of the formula.
Especially since the problem is being framed in a way that OP thinks is incorrect and misleading.
I don’t like “back channels” on open source projects because they make it harder for new users to get into the project. Discussions should be open and accurate, not behind closed doors.
As a potential donor to this project, I’m glad that OP posted this as otherwise I wouldn’t know about their weird way of dividing donations.
It's pretty amazing how useless twitter threads can be for actually containing the relevant information, even if it has actually been tweeted as part of it. He has more specific allegations if you dig around his twitter profile for a bit and you can in principle look most of the details up because the development is all public. TBH sounds like this could become a bit of a mess, if it isn't already.
Yeah that's a big claim to make, from my understanding his point is the money is misallocted while there are bunch of people putting more effort on the project who doesn't gets paid or getting paid lower then Henry. (Based on commits, pull requests, issues etc.).
If I don’t have an employer and instead get an autodonation without any supervision, then that’s the donor’s problem. Or teammates if it affects the level of future donations.
> Babel used by millions, so why are we running out of money? Bluntly: Because funds were misallocated for years, and the project has been too slow to improve.
> The reason there's no money is because someone took a $130k annual salary and didn't actually work on the project.
> I am sitting on an ivory tower so I know this is a terrible look. The elephant in the room is being ignored and it's hard to see so much sympathy for a self-inflicted problem that isn't being honestly publicly addressed.
If they worked, would Babel have more money? Looking at their website, it seems it is a free project. No mention of any paid product. Then how would they have more money had one person worked?
Look at another aspect. There's the work aspect: the lucky fellow who was paid $11k/month for creating one PR and two issues. There's also management/overfiew aspect: the organisation that didn't notice that it was paying.
If the organisation (I realise it's just a few people) were paying enough attet6ion to notice that it was paying for ~nothing, then it's a fair assumption that it would also be paying enough attention to its sponsors.
Definitely it is very surprising that they continued to pay despite there being close to zero output. If I tried this in my company I'd have a PIP in 2 months and probably get fired in another 2.
They could have more money for 2 reasons: 1. People may have stopped the support seeing how the existing funding is mismanaged. 2. People may not support seeing how little work is done with the existing funding.
Basically better state of the project = more likely funding.
The "open-source funding crisis" is simply the crisis of not knowing how to reliably organize our society in any way other than for-profit companies but wanting to direct our useful energies at something other than propping up those for-profit companies.
I get paid well over $130K a year, and if I manage to get 12 issues, 25 comments, and 29 pull requests to open source projects, that's a good year. Where does my salary come from? What do I actually do with the rest of my time? The world has no visibility into that. (The answer is - a hedge fund. Turns out there is no hedge fund software engineer funding crisis.)
There's nothing wrong with getting paid $130K a year to not work on open source. There's specifically nothing wrong with getting paid $130K a year from the organizations who you want to be funding open source. Babel's highest sponsorship tier (https://opencollective.com/babel), at $24K a year each, comes from Airbnb, AMP (Google), Trivago, and Adobe. Other top contributors listed on that page include Salesforce and Facebook. Do we complain when those organizations spend well over $130K a year on not-open-source? Do we complain when those organizations spend well over $130K on internal projects that just completely fail?
I'm starting to think the answer is that we should, in fact, complain. As the linked Twitter thread points out, Babel isn't running out of money because $130K is too high a salary - it's running out of money because the lack of visible results endangers their ability to get future funding. But the amount of funding they're getting is scraps from any of these organizations that, I'm sure, get way more than $24K/year value from Babel. If you have a thousand software engineers, and a piece of software makes them all just 1% more productive, its value to you is ten full-time developer salaries.
And all these companies are more than happy to employ thousands of software engineers, plus hundreds of managers and other human infrastructure to support those engineers. And they understand that sometimes those engineers don't deliver value - usually because they were told to work on the wrong thing - and that's a correctable problem and you keep paying those engineers while you correct it. If they have a question about whether someone is doing the right work (maybe they're doing good work, but it's not what was agreed on paper - it certainly seems like Henry is doing a lot of code reviews and publicity work?), having skilled managers is how you figure out whether that work is nonetheless worthwhile, or the engineer needs to be told to shift their focus. We understand that having managers, who are also paid more than $130K/year, is important to a functioning engineering organization. (Usually they're traditional line managers, but even at organizations that reject traditional managers, you still have senior contributors whose value is primarily in how they guide and advise other people's work and not so much in their own work.)
It may be true that an open-source project's future is in danger because they've employed one person and that one person isn't visibly productive, but the mistake here is on the expectation that this could ever work. Imagine if Adobe employed one engineer to work on Photoshop, discovered they were not terribly productive in terms of new features implemented in Photoshop, and then worried because they'd have trouble begging the CFO for funds for a second engineer. We would see this as ridiculous. If you want to see continued development on Photoshop, you need a team. If the senior management of a tech-driven organization refuses to proactively staff an engineering team with more people than strictly needed, no one would think to beg. We'd just know this management is incompetent and short-sighted and not worth bargaining with.
(And, like Babel, Photoshop has been around for many years and does its job just fine. In a sense it doesn't need new features. But I'm sure that Adobe has way more than one engineer on the project, nonetheless.)
We need to figure out how to stop begging for scraps from companies that regularly fund hundreds of millions of dollars in development for code that only benefits themselves, and start saying, look, this open-source project brings you millions of dollars in business value per year. $24K is a joke. We're not going to try to justify the additional value of a second contributor, when you've got ten highly-paid people on the committee deciding whether to give $12K or $24K. Give us millions of dollars, then you might see some results.
>But the amount of funding they're getting is scraps from any of these organizations that, I'm sure, get way more than $24K/year value from Babel
It's hard to measure the value it provides, but generally thats a weird way to look at things that are open sourced. If they wanted their fair share of the profits, they could've chosen a different license, or sold the product.
I'm not talking about fair share of the profits - the OSS project, I hope, has no profits.
Babel, in particular, is a project that's primarily interesting to software development companies. It's not consumer-facing software like Signal or Firefox or GNOME or whatever. It's not the logic for a self-driving car or a pacemaker. It's a tool that's for software developers who want to write in one version of JavaScript and have code run in an environment that expects another version of JavaScript.
I would agree with you for things like Signal or Firefox - that while there's theoretically a monetary value for me being able to use them, it's difficult and weird to pin down that number, and it's a social good to have them available for free.
But there's a very non-theoretical monetary value for companies that use Babel. They can develop software x% faster, and they pay $y per unit time to develop software, so the value is x times y. It doesn't even matter if the company using Babel makes any profit: their costs are lower because they use it.
And, corresponding, Babel isn't a social good like Signal or Firefox. It doesn't make the world a better place on its own. It makes the world a better place insofar as people using Babel are trying to make the world a better place and they can do so faster. (Or it makes the world a worse place if people using Babel are trying to make the world a worse place.)
So it's up to the people using Babel, all of whom have significant engineering funding, to make sure it continues to exist and work. And they should be doing so on a scale commensurate with the cost savings they get from Babel. Which is to say, they will do so, provided that the danger of Babel not existing is real, and the possibility of avoiding that fate with funding is real - it's in their rational self-interest to do so.
Or, in other words, the framing of "funding sufficient to keep a project alive" as "fair share of the profits," and the idea that closed-source software companies can reasonably expect "profits" and open-source projects used by those companies cannot, is itself the crisis. We have no way of talking about funding apart from profit. Either we need to come up with one, or we need to admit that the only way to solve the open-source funding crisis is giving up on open source.
If we can't figure out how to employ more than one person on Babel when its users' rational self-interest would be to fund it to the tune of many millions of dollars a year if they had to, is it any surprise Signal is pivoting to cryptocurrencies and Firefox is pivoting to VPNs?
At a previous company we got rid of Babel on our node stack completely. Felt good. Not possible yet with web but that is what we should be getting to. Funding Babel for the indefinite future means the web will have stagnated. In an ideal world Babel dies.
Babel doesn't necessarily need to transpile all the way to ES5, you can specify targets. It's a useful tool for using newer features even without supporting IE
The place where I see it as still necessary is transpiling code that uses es6 modules for web. If browsers all supported es6 modules and not half heartedly I think myself and a lot of devs could ditch Babel for good.
But it is useful for general signs of activity. So I think it’s stupid to say stuff like “foo has 120 commits and bar has 119; therefore foo is more productive”
But I think it’s a good starting point to think something like “Hrm, Bar only has 21 commits in a year and the rest of the team has 120. Let’s look into that.”
Sometimes there’s a good explanation (eg, “Foo makes a huge commit every two weeks, let’s talk to them about the benefit of daily check ins and branches”) but it also helps identify real problems.
I once worked with a company that was wrapping up a project. They have contractors 30 days notice to finish their work at the end of the release and go. One contractor went from regular activity to zero commits during the last month. Since I was supposed to integrate and code review his stuff, I asked if he would check in and he just said “yeah sure.” He left with no commits during the month. Not that there’s a specific number of commits that was good and he was obviously mentally checked out, but zero indicates something worth investigating.