Note: Im purposely ignoring the pyramid scheme like effects from rising currency values.
2k per engineer ... that's more than a rounding error. It'd be interesting to compare to per-seat licensing for commercial software. Anyone have examples?
For a free and open source project, the ultimate goal IMO is growth and usability, to fulfill the reason why the software itself exists, and that can be done by donating that money in the shape of a hiring dedicated to work on the software. Some companies already do this, and i don't see why the practice couldn't be more established as a way to ensure a quid pro quo with the community.
A couple of years back, someone asked if they could donate $250 to me - that was the first (and until now, only) time anyone did such a thing. I was really happy about it - not so much about the money, but more the fact they liked the project so much they were willing to hand over money.
Anyway, I asked if they could donate it to a charity instead, and I suggested a few that have personal meaning. They were quite happy with that, and actually donated $500 instead :)
Plus there's no guarantee that the developers will be the ones making money. There's a lot of ire at Amazon because of that.
True though that for larger projects there can be a lack of ownership problem. The conversation could become "hey we'd like to lend you our legal resources to help you set up a foundation for this project, and then we'd like to donate $X/month to the foundation going forward." When it comes down to it, if a project is able to cut releases, there is some sort of structure in there that is working, just have to figure it out.
An easier way to scale this might be to just have a set amount (say $10k/month for a large company) and a recurring ranked-choice poll open to all employees. The funds would be allocated according to the poll, with the winner getting say $5k and the two runner ups getting $3k and $2k or something like that.
This way employees don't have to do the work of submitting donations and filing expenses, as there's just one person at the company who administrates it a few days a month.
I think this is the right insight, but clicking around the links on their website I don't think they were executing on that insight particularly well. This looks more like it was a platform for individuals to make recurring donations than for companies to support the tools they use.
Probably why they don't exist anymore.
https://liberapay.com/ a different team/company took over the assets.
The only way this will ever get resolved is if companies collectively join together and decide to pay. Like say, we could have a tax, that they pay the country they operate in, and that tax could then subsidize these free open source projects that produce positive externalities.
I work with a few companies with ~50 tech employees each. Both use significant commercial tooling. Both pay far less than $100k a year for it. If their use of open source (some scattered linux, but everyone has also a Windows PC and/or a Mac, etc.) cost them $100k a year, they'd both drop all of it and move to cheaper commercial solutions.
Pick a field. Pick commercial tools that a technical person in the field uses. Check prices spread over the lifetime of the tool.
Not even close.
Art people - $2k/yr buys a lot of solid software that open source doesn't come close to.
Software - $2k a year buys a significant amount of commercial tooling.
CAD - Open source doesn't even have an answer here.
Finance - Open source sucks here too.
Scientific computing - Matlab/Mathematica/etc - open source not close, $2k a year buys all your tooling over the lifetime of the tools.
And so on....
The reason I use some open source solutions is because they're adequate for the needs, and the cost reflects it. Had I needed to pay $2k a year for them, I'd have dropped them and bought decent commercial alternatives.
At Arist (YC S20) we donate $1k/month to the main open source project we make use of (Ruby on Jets). This has been a major productivity boost for us as issues critical to our business are fixed quickly. Definitely recommend it!
Having in-house contributors spend company time on these dependencies alleviates the issue.
The reality is, if these tools are so worth it, the company will support them in house as needed, but the idea that community needs to pay for software suddenly because some people want to work on it full time is a bit odd.
Encouraging more volunteer time is better than trying to collect and allocate cash. And better improved OSS. I think.
There’s different motivators for different people. I would not want to pay open source developers and I would rather pay commercial companies.
Of course, I would rather contribute back to the open source community through my own contributions as this leads to a stronger community.
If there’s a transactional aspect to OSS then it could make the culture worse in the long run.
The article proposes that per person not per company.
> Encouraging more volunteer time is better than trying to collect and allocate cash. And better improved OSS. I think.
Depends on the size of the company and the number of projects you want to support. A bunch of low level commits across a bunch of projects by 100 companies would leave things off a lot worse (due to the maintenance burden vs improvement amount provided) than a bunch of donations from 100 companies to get a dedicated resource dedicated to each project year round.
During heavy periods this year I had ~20-30 code reviews per day.
There's a balance between money and contributions. You can't just throw bodies at a problem and say it's better than any alternatives.
I think that contributions can be done in a smart manner that doesn’t overwhelm maintainers.
Mostly I think the expectation is that companies should sponsor more open source work.
* There's nothing about who is doing this or why.
* There's no way to actually pay that I could find.
* There's no information about overhead, or if the GratiPay is pocketing the money.
It's just a half-baked half-finished web page. I think a good not-for-profit here might be helpful? But I'm not sure how to structure it so that it is. I don't see why a business would actually want to pay. Most managers believe they have a fiduciary duty to maximize shareholder value. This doesn't help.
And the money raised? I'm not sure a serious analysis has been done by anyone. As an open source author, I'm not inclined to even deposit a $0.32 royalty check.
“Gratipay: A pioneer in open source stability née gittip 2012–2017 RIP”
Apparently, after five years of trying to make a business of it, he gave up.
The article, by the way is from 2017. Perhaps a mod should add that to the link title.
I've been doing this myself with a combo of one-off donations (Ubuntu, Pop OS, libinput) plus experimenting with Github Sponsors on one project.
Here's an example from TripleByte of their employee-directed donations: https://triplebyte.com/open-source-donations
One of the reasons why devs at enterprise scale companies like OSS so much (except GPL, obviously) is because trying to purchase anything is a nightmare.
If you've worked at such an org, you know exactly what I mean, otherwise you might even find it difficult to believe. Purchasing anything can take several months, with hundreds of manhours wasted on call, meetings, legal review, approvals from random people etc. And if you're the one that initiated it, you'll be the poor sod that has to keep pushing that boulder up the hill all the time, otherwise it just stalls forever. It's honestly should destroying.
Businesses teach accounting and finance, very numbers-oriented subjects, as a minimum, so this is not entirely outside of the wheelhouse of business schools.
And yet this near-universal problem of software procurement continues across enterprises, a problem traced basically to equally universal teachings on cost control policies and other nuts and bolts that come out of explicit teachings of b-schools and the general literature of running businesses.
Since I posted a wandering treatise on american antiintellectualism and the fundamental lack of cool of the "nerd", and how b-schools want to appear as cool, cutting edge producers of captains of industry (that adopt learning nerdy accounting as a necessary hurdle to get as much money as possible), I doubt they will properly pivot anytime soon.
And who is doing the support?
It sounds like you are envisioning something like Red Hat.
Note that you can have a one week (or more) lead time for support requests: the timing doesn't matter, you only need a maximum window so they can share visibility and socialize issues like this: "ticket opened with vendor, next update in one week".
Of your five examples, the first two reinforce $2k/yr as a reasonable spend, the second two are no-ops since you don't give a number, and the fifth case is unclear to me (a quick check of Matlab shows a $2k perpetual license with some fine print about support contracts, or an $860/yr annual license).
To your secondary point: the fact that in some fields there are no good open source alternatives suggests that there's room for open source alternatives to exist, if our companies can become self-enlightened enough to fund them. Yes, there will always be companies that mooch. I, for one, prefer to work for companies that don't. :)
How so? Annually? Nowhere I know spends that per employee in (software) tooling.
Here's the first 2 you mention as reinforcing $2k as reasonable spend:
Art people - Photoshop creative 600/yr, and there is nothing close to it in opensource at any price. Many creatives don't need all of ps, only a piece or two, much cheaper. OS is order of $100, lasts 5 years easily. Or maybe they're doing high end video editing in DaVinci Resolve, again with nothing even close in open source, for $295, again good for many years. Some artists will use a few tools, but then again many places also have floating licenses.
Software dev: OS again $100, lasts 5 years easily. VS Pro 2019 $499, again many places keep a version about 5 years. So far we're at $120/yr for a significant amount of dev tooling. Again, some people use multiple paid for tools (say Resharper, etc.) but many don't. And again, lots of commercial tools have floating licenses to lower costs.
Care to detail an annual software spend approaching $2k for either of these fields you claim provides reinforcement for that amount being a reasonable spend? I'd be curious how you spend $2k/yr for the average developer using commercial tooling.
>the fact that in some fields there are no good open source alternatives suggests that there's room for open source alternatives to exist
Actually, it points to open source being unable to meet market demand. That there is no CAD system comparable to a professional CAD tool, despite there being CAD tools for over 60 years, and despite it being a huge market, shows that open source simply cannot compete for many markets. This is true is nearly all products except a tiny few: Linux is a good OS, a few good web stacks, thunderbird, etc. But it sucks for the vast majority of product categories used by professionals.
For CAD, basic SolidWorks is $4K, is insanely powerful, and a version easily lasts many, many years. And it has floating licenses.
>Yes, there will always be companies that mooch. I, for one, prefer to work for companies that don't.
I prefer to work for companies that apply the best tools for the job. If the tools are open source, so be it. If they're closed source, so be it.
It's fine to raise money to fund open-source, it's fine to make closed source and sell it. Offering something for free, then calling it mooching when someone uses it for free, is simply childish emotional blackmail.
zero negotiations, zero lawyers, just a commandline one-liner or something.
micropayments for package management.
Funny timing, btw. :)
https://twitter.com/chadwhitacre_/status/1431368348886052865
> Open source doesn't even have an answer here.
> Open source sucks here too.
> open source not close
> But it sucks for the vast majority of product categories used by professionals.
This is all irrelevant to me. Companies should pay for the open source software they use today, not for the bad and non-existent alternatives that they don't.
> Offering something for free, then calling it mooching when someone uses it for free, is simply childish emotional blackmail.
Meh. As my kids would say, "That sounds like a you problem." :D
Just because open source maintainers legally allow companies to mooch doesn't mean they're not mooching. ;)
But I mean don't get me wrong, maintainers can be a whiny bunch sometimes for sure. ;)
> open source being unable to meet market demand
Open source software is a public good, and markets are so bad at efficiently provisioning public goods that this is in fact their text-book failure mode:
https://www.investopedia.com/terms/m/marketfailure.asp
In other words, _of course_ Photoshop trounces GIMP. Use Photoshop, by all means! :D
But! What if we can evolve our markets? That's the macro opportunity here:
1. A critical mass of companies voluntarily pay for the open source software they use today. This is primarily in developer tooling as you point out (Linux, web frameworks, etc.).
2. Existing open source projects become more professional, by which I mean they get better at things besides coding, like design, sales, support, management, etc.
3. Software other than developer tools starts to be built as open source. This depends on the first two things happening.
Why is this interesting? To me it's interesting because it's a slight glimmer of a real opportunity to decouple work from employment, so that individuals can work collaboratively more out of intrinsic motivation than for a paycheck. This doesn't solve all the world's problems and doesn't come without its own challenges, but I think it's an interesting and generally positive evolution of society that is fun to be a part of. :)
[edit - formatting]
My statement was based on these two statements of yours:
> Art - $2k/yr buys a lot of solid software
> Software - $2k a year buys a significant amount of commercial tooling
I read that and thought, "Okay, well, I want a lot of solid software, so $2k/yr ... check!" I guess that's not what you intended. :)
> Care to detail an annual software spend approaching $2k for either of these fields
In any field, the approach I would take is to think about the SaaS products that a professional uses, because that's where the bulk of the software spend is these days. In software dev we have things like GitHub, Bitbucket, Datadog, New Relic, Splunk, Logrocket, AWS, GCP, Vercel, Figma, Canva, Slack, Discord.
I spot-checked GitHub, and their enterprise tier is $21/user/month or $252/user/yr:
Slack is $150/user/yr for business, so enterprise is probably 2x that ~= $300/user/yr:
Seems like at Salesforce $300/yr is the low end, and they go up to $3600/user/yr(!):
https://tech.co/crm-software/salesforce-pricing-how-much-doe...
Let's say the average enterprise SaaS login is $250 or $300/yr. Six or eight of those gets us to $2k/user/yr, and six or eight SaaS logins seems pretty reasonable to me.
Note that in the OP blog post the 2k/user/yr number is arrived at from the top down, not from the bottom up as we're doing here. What I'm interested in in this thread is cross-checking the order of magnitude ... so far this thread confirms for me that 2k/user/yr is probably in the right ballpark for large companies to spend on their open source dependencies.
It doesn't sound like you or I have the expertise to go much further down this line of questioning. I'm sure people who control IT budgets at large companies have visibility into their average annual per-user software spend. It would be interesting to hear their view. :^)
[edit - formatting]
> GitHub, and their enterprise tier is $21/user/month
Github has a $4/month pricing plan. The companies I deal with all use internal source control, paying zero to github.
>Slack is $150/user/yr for business,
Of the 4 corporate slack channels I am art of, only one is paid, using the 6.67/mo/user version, for $80/person/yr. So on average these 4 run $20/user/year.
Want more data? Here's [1] Stackoverflow 2020 developer report. Half report using Slack. If that are like my experience, then maybe 1 in 4 uses paying, and maybe that's the low tier. If that still holds, 1 in 8 devs from StackOverflow (which is a very select group) is paying for Slack.
>Salesforce
Wait, what? Are you claiming every dev needs Salesforce? I know precisely zero with it, and I run a decent amount of seminars across multiple companies. Are you a developer that uses Salesforce regularly? This makes no sense to me whatsoever.
And then you throw in some vague SaaS products, needing to get to 6 or 8? Go ahead and tell me which 6-6 Salesforce style SaaS things the average dev uses.... It will be interesting to see what you pick.
So sorry, it seems you're more than just making stuff up, picking the highest values at each choice, to try to justify the prices.
>It doesn't sound like you or I have the expertise to go much further down this line of questioning.
I work directly under the owner and VP of tech in one company, for which I've been doing so for almost two decades, and have known a decent amount of their their finances for a long time. I also have talked at length with the Pres, VP, and head of HR at the other company I regularly work with. I own a profitable company for over 15 years now. All of these are heavily software dev houses. I have a decent number of friends that own dev companies or are high enough up to know the answer to this, and I've regularly asked them how their internal cost structure breaks down, mostly to compare to mine and to refine my understanding of how company finances work. I'm pretty sure I have decent insight to how this works.
For the first I write the proposals to get new software in, I see the counts, I price options, for a large amount of the developers. I see the numbers there.
At the second, from talking to those who do exactly this, I am quite clear their rates and overhead is in line with what I expect from such companies. I could ask directly, but since I work with a lot of devs there and see the tooling, there's no need. $2k/dev/year is a ton of money.
At the third, since it's my company, I see every nickel. It's also no where $2k/dev/year. Buying hardware like 3d printers, laser cutters, PCs, etc., ups the per person spend, but even there it's amortized over people and time and still not likely 2k/person/year spend.
It sounds like I do have expertise in this area and you are missing it.
[1] https://insights.stackoverflow.com/survey/2020#most-popular-...
Hope the option package is worth it.
Some examples of this, that I can think of:
- Varnish Cache where you need a paid subscription, or community patches and you have to do the compilation yourself, to make it understand SRV-records instead of hardcoding IPs or hostnames.
- Nginx also need a paid subscription to understand SRV-records.
Both of these companies have realized that in order for people to effectively run the software within a cluster, you need support for SRV-records to get service discovery working.
I can imagine it would include things like an office space, snacks, yoga classes, etc. I don’t think it is anywhere close to additional $150k, maybe $10-20k.
Anecdotally, $2,000 per user per year would be a really expensive SaaS app (right?). Not exactly apples to apples but so far I think $2,000/user/yr is a good ballpark for open source value. Note also that the blog post indicates a 50% markdown in labor value because companies pay a tax to direct employee time, so people should expect to get paid less when working self-determinedly on open source.
https://wintelguy.com/oracle-db-licensing-calc.pl
https://www.orskl.com/what-is-the-cost-of-oracle-database-li...
Not seeing you at github.com/kawsper so not sure what the PR was ... but thanks for the PR! :)