How to trade software for small money?
scattered-thoughts.net
scattered-thoughts.net
I find this an interesting development, to be honest. People were outraged over Redis going "free unless you're a business", but the people who were affected the most were the people who have profited off Redis for ages. Every time a free product goes "pay us if we make you money", people get angry, but I rarely see the upset people start a fork with the old license.
I think the fact we're so used to "just" downloading dependency upon dependency plays a role. If one dependency asks five dollars per year, that's a no-brainer, but if the hundreds of dependencies that make up the modern software stack do, that becomes a problem.
I wish more people would buy and sell GPL/dual licensed software. GPL for the software freedom that open source once was all about, dual licensing for the companies that don't want to give their customers the freedom that comes with the GPL.
On the other hand, I also feel that switching to a profit oriented work isn't acceptable if you've accepted sizable contributions from the community. This is legally acceptable with a CLA or a license that permits most commercial use, but ethically difficult, unless you also pull every major contributor in with your licensing-for-money scheme.
In the end, I feel a lot of the protests and anger about selling (partially) open software comes from a sense of entitlement. I know "fork it if you don't like it" feels like a cop-out, but with how few maintained forks I see of software that caused controversy, I feel like a lot of this sentiment is people wanting other people to maintain important parts of their software for them.
I think it's important to understand that Redis as a whole is more than just Redis the product. It's also the social and intellectual capital invested into using Redis from the community and includes every StackOverflow answer and anyone whose resume has Redis on it. If Redis didn't exist, teams would have chosen any of the other alternatives: Memcached for caching, Kafka for queueing, etc. Redis positioned itself as the conceptual floor tile of "free thing to use for X" and then decided to move itself somewhere else, and if you were standing on the floor in that spot and happened to be impacted by the licensing change then you now found yourself hurtling to the ground, time wasted.
This can feel -- to some people, at some times -- like a cynical rug-pull. Developers get intangible but real benefits from creating well-known software, and open source software becomes well-known more easily. Changing the license once you've extracted those benefits means that the effort put into learning to use your software and into writing tutorials, Q&A answers, and conference talks was partly wasted. Attempting to capture part of the pie means reducing the size of the pie: across broader society, value is lost as the positive externalities created by your product shrink.
I do think that in the specific case of the SSPL the argument is more complicated and I tend to side with Redis Labs more than it sounds like in the rest of this reply. I just disagree with the idea that there is always zero problem with relicensing something you have rights to.
I agree with this analysis, but I think the conclusion to draw from it is not that this is an unfair outcome for the poor person standing in that part of the floor, but rather that this is an unfortunate outcome that that person should have considered likely to happen from the outset, and prepared for. (I'm assuming here that no promises to remain FOSS forever were actually made.)
Making a good thing available to people ought not to shift onus onto the good-thing provider to continue providing it forever.
It sucks to see an open source project become proprietary through the SPSS, but if the community is as strong as all of those social and intellectual investments make it seem, there will be no problem maintaining an open fork. I imagine the independent open source contributors won't be as willing to contribute to Redis now, so there's a chance to maintain the open source version.
This approach worked for MariaDB, LibreOffice, and a bunch of other projects that went through similar controversies.
A central reason was that the SSPL (license) is not an open source license (as defined by the OSI: https://opensource.org/osd ). See https://opensource.org/blog/the-sspl-is-not-an-open-source-l... for details.
This means that Redis switched away from being open source.
The state of open source is really unfair to maintainer.
The more popular a project becomes, the most cost is involved in maintaining and supporting it, but all too often (like with redis, elastic search, and aseprite) there is 1 company or even just 1 person facing the brunt of the community and development burden.
Building a community for your open source project is a whole job in and of itself.
I love open source. I owe my career, hobby, and lifestyle to it, but the cracks are showing more and more. As a OSS community, we need to be more active in supporting the project we use for free otherwise what happened with redis is going to be the end state for many very popular OSS projects.
If the industry comes together, and Redis only sells limited copies, as I would expect from a previously free product, it will effectively look like Redis is a fork of its open source version, not the other way around. If the industry doesn't (because that would require time, money, and effort), I can't say I have much of a problem with the situation, because apparently people were using the Redis devs for free work and not much else.
- open source works because there is a large amount of work that's done by people other than the main authors, or people outside the project. e.g. See all the documentation that's done around Linux or git.
If you try to sell a small piece of software, I believe in most cases that the cost of developing will be dwarfed by the cost of support, documentation, and marketing.
- Giving away almost everything is standard for both open source and for for-profit.
That seems to be inherent in the structure of the economy at this point. If you don't give away a lot, you don't really compete. e.g. see the social media companies that wanted to charge for usage
sqlite seems to have sustainably paid employees, but it gives away 99.9% of the work. Similarly, free tiers often give away 99% of the features of commercial products, with some artificial caps.
- Likewise, going into debt seems to be standard practice. It's hard to be competitive against organizations taking on huge amounts of debt.
You might not think it's the same at a small scale, but I do believe there are say lots of crypto companies hiring technical staff that "distort" the market of who can help you, e.g. compiler engineers
- SaaS seems to work more at the application level. I always thought it would be nice to sell low level services like data feeds, but in practice that seems rare. (I guess Bloomberg terminal is kinda like selling data behind an expensive piece of hardware. Similar to Apple, if you have the resources to achieve this vertical integration, then you can have a "moat")
---
It would be nice if there were other models for sure. But I don't think the problem is necessarily software, as much as "small".
I think the economy is structured in a way so that the competitive actors tend to be groups of people aligned in some corporate structure (which does kinda suck). As in the beginning of the post, it says "your competition is highly paid teams are profitable companies".
Even a big company like Hashicorp has problems competing against those companies, and apparently Redis. And also nytimes, Universal music, and every other media company. The economy is not a friendly place.
Certainly there are one person businesses like tarsnap and pinboard and so forth, but I also think marketing and support is a big problem in those cases.
I don't get your point what crypto companies have to do with compiler engineers. To my understanding, the programming skills that a crypto company needs are quite different from compiler development.
Ethereum, ICP, and other layer 1s include compilers to turn smart contracts into bytecode that can run on their respective networks.
Randomcoin companies don’t need compiler engineers, I don’t think.
On a purely personal level I still haven't found a better model than this. (That speaks poorly of me, not well of the model.)