Large Corporations Aren’t Going to Save the Real Linux Community(techrights.org)
techrights.org
Large Corporations Aren’t Going to Save the Real Linux Community
http://techrights.org/2020/11/08/faking-communities/
70 comments
Open source software is dying because software is dying. Once upon a time people lines up for hours to get the latest software. Boxes were and CDs were proudly displayed on office shelves. But software isn't the hot thing. Services are king. Facebook and youtube aren't software. TikTok isn't software. They are services and services are much harder to open source.
There are a few. Tor and Signal are good examples. They are software that enables an underlying service. So they get attention. People want to contribute. Imho if F/OSS wants to thrive it has to focus on those projects that provide services that people can proudly say that they use and to which they can proudly contribute time and energy.
There are a few. Tor and Signal are good examples. They are software that enables an underlying service. So they get attention. People want to contribute. Imho if F/OSS wants to thrive it has to focus on those projects that provide services that people can proudly say that they use and to which they can proudly contribute time and energy.
Thank you for bringing up a new angle that I hadn't considered before :)
Yes, we should look into how one can adapt the "free speech" open source spirit to work for services.
Yes, we should look into how one can adapt the "free speech" open source spirit to work for services.
Self-hosting. Otherwise you’re forever limited to whatever the service provider’s understaffed moderation team translated from the corporate email and plugged into the filters this afternoon.
While self-hosting is a good option for us developers, I'm pretty sure that most normal people will not do it. So to reach the critical mass needed to make an open source project financially viable, I don't think we can rely on self-hosting.
As an addendum, this is also where (extended-family-level or city-wide) community-hosting comes in though unless your services are federated or have other attractors, it's difficult to attract users.
You’re not wrong, but since service providers control what happens on their servers, and have no obligation to allow any particular speech, then self-hosting (or its close cousin, federation) are the only ways to assert publication control of your own posts.
> look into how one can adapt the "free speech" open source spirit to work for services
Unless you want to treat service providers as public utilities, which isn’t necessarily a bad idea, just an uphill battle.
> look into how one can adapt the "free speech" open source spirit to work for services
Unless you want to treat service providers as public utilities, which isn’t necessarily a bad idea, just an uphill battle.
> Signal are good examples.
Is Signal a good example? It is not federated and you can't connect to an alternative server from the Signal client. When I use signal I feel like I am using a free service.
Is Signal a good example? It is not federated and you can't connect to an alternative server from the Signal client. When I use signal I feel like I am using a free service.
well there are many services people use proudly. Previously authors of opensource projects used to run consultancy for their softwares. But now days, many such projects get ported to public cloud in no time. So for these projects, public cloud providers are only (direct) customers.
IT IS TIME TO UPDATE YOUR LISCENSE TO ADDRESS PUBLIC CLOUD PROVIDERS else sooner or later open souce gonna die
IT IS TIME TO UPDATE YOUR LISCENSE TO ADDRESS PUBLIC CLOUD PROVIDERS else sooner or later open souce gonna die
We seem to be slowly moving from an ownership model to a subscription model where you pay "rent" with money or data. Not just software but hardware as well. How long before smartphones are simply fused shut preventing you from opening it and you have to return the smartphone before upgrading to another.
That's a somewhat consumer-centric view of the software world. But it's true that, even in business, there is something of a shift towards managed services vs. on-prem software.
> The GNU project makes it very clear that they are striving for the freedom of the user to modify and adapt the software that they use. So every time a GNU tool is used in the private back-end part of a SaaS webapp, that's directly opposite to the spirit of the GNU project, because you, the end user, have no way of modifying said SaaS.
> Also, I would be very hesitant to release anything valuable with LGPL or even GPL these days.
The point of the AGPL is to close the loophole that allows this abuse. Just switch to it for your own projects.
> Also, I would be very hesitant to release anything valuable with LGPL or even GPL these days.
The point of the AGPL is to close the loophole that allows this abuse. Just switch to it for your own projects.
As far as I understand, the AGPL still allows a service provider to use an unmodified binary internally, as long as the user of that service never accesses it directly. If my understanding is correct, that means if GNU were to switch to AGPL, it wouldn't help the users of SaaS webapps that internally use GNU tools at all.
The point of the AGPL is to require sharing of modifications to the software when that software is used in a service. If you're using an unmodified binary, then you haven't made any changes to the software. You won't be obligated to share changes if the changes don't exist, so you could use the unmodified binary internally or externally.
The AGPLv3 has a lot to say about patents, which is enough to scare off many companies from software that uses the AGPLv3, but I believe it won't be triggered if you never make any changes in the first place (and are therefore presumably not a 'contributor').
AGPLv3: https://www.gnu.org/licenses/agpl-3.0.en.html
See also https://en.wikipedia.org/wiki/Affero_General_Public_License , https://softwareengineering.stackexchange.com/questions/1078...
The AGPLv3 has a lot to say about patents, which is enough to scare off many companies from software that uses the AGPLv3, but I believe it won't be triggered if you never make any changes in the first place (and are therefore presumably not a 'contributor').
AGPLv3: https://www.gnu.org/licenses/agpl-3.0.en.html
See also https://en.wikipedia.org/wiki/Affero_General_Public_License , https://softwareengineering.stackexchange.com/questions/1078...
If the webapps themselves use AGPL software, that's not "internally." If the employees of a SaaS company themselves use AGPL software that isn't linked to the service that users access, that's internally; e.g. if a webapp uses an AGPL database, the users are accessing that database.
I'd say using an AGPL tool as a service for your webapp is this scenario:
"Scenario 1: Using an unmodified AGPL binary" https://medium.com/swlh/understanding-the-agpl-the-most-misu...
which seems to be allowed by AGPL without disclosing source code.
"Scenario 1: Using an unmodified AGPL binary" https://medium.com/swlh/understanding-the-agpl-the-most-misu...
which seems to be allowed by AGPL without disclosing source code.
The details are somewhat unclear but basically yes, that's what a cloud provider/SaaS company would probably assume. So they probably just won't use AGPL code and if they do want to offer a service with the same APIs, they can (at least for now) recreate a compatible service that they write themselves.
Just myself, I saw an increase in FOSS activity seemingly due to the Great Recession leaving so many unemployed. People had so much time on their hands that contributing to a FOSS project made sense for all sorts of reasons. After years of recovery, people are (were?) back employed and don’t have such time on their hands.
FOSS is pretty interesting as a way for otherwise non-cooperatives to make something useful together. Also seems like something similar to academics. Unfortunately, our economic system might force it to be a fluke.
FOSS is pretty interesting as a way for otherwise non-cooperatives to make something useful together. Also seems like something similar to academics. Unfortunately, our economic system might force it to be a fluke.
Yes I agree with this, and in reality nothing really stopping Companies from abusing the GPL, notice how almost all corporations avoid GPLv3 like the plague.
(edit, I meant easy to abuse GPL2, but GPL3 ias hard to abuse)
Personality I believe the final nail was the "coup" at the FSF with RMS. He could be a pain at times, but almost all his fears about the future of computing is slowly coming true.
(edit, I meant easy to abuse GPL2, but GPL3 ias hard to abuse)
Personality I believe the final nail was the "coup" at the FSF with RMS. He could be a pain at times, but almost all his fears about the future of computing is slowly coming true.
I'd say GPLv3 avoidance demonstrates respect[fear] of the license, not abuse of it. To abuse it would be to use that software while not respecting the terms of the license.
I missed pointing out GPL2 is rather easy to abuse, i edited my original post.
Yes, Large Companies is scared of the GPL3 and we are told specifically never use any GPL3 software where I work :)
Yes, Large Companies is scared of the GPL3 and we are told specifically never use any GPL3 software where I work :)
Like CISCO, Gigabyte, D-Link, NETGEAR, AVM, FANTEC, Iliad, ...
https://gpl-violations.org/news/
https://gpl-violations.org/news/
GPLv2 vs. v3 seems to be more a matter of inertia. Yes, a few have concerns about the TiVo-ization language. But a lot of it is that projects like Linux were already under GPLv2 and Linus wasn't interested in changing--and, absent copyright assignment, doing so would have been at least somewhat controversial and would necessarily rely on legal theories not everyone would have been on board with.
RMS was right and the community is afraid to admit it
> "It ruins my mood if I get an entitled email from some random stranger who is insulting me because my free app did not solve his/her problem [...] apparently, re-selling PostgreSQL as a service with presumably millions in revenue is not enough for Microsoft to become an official sponsor."
Aren't these part of the freedoms of speech and behaviour you were enthusing about a couple of sentences ago? And your freedom to stay away from open source is part of it, too?
"Open source with an ideal of freedom, where the users (corporate and personal) behave exactly how you want" sounds contradictory.
(Curious, are you happy that Microsoft is part of the Linux Foundation? Or do you see that as them exerting unwanted influence?)
Aren't these part of the freedoms of speech and behaviour you were enthusing about a couple of sentences ago? And your freedom to stay away from open source is part of it, too?
"Open source with an ideal of freedom, where the users (corporate and personal) behave exactly how you want" sounds contradictory.
(Curious, are you happy that Microsoft is part of the Linux Foundation? Or do you see that as them exerting unwanted influence?)
Seems unnecessarily pessimistic. Just because there are a few rude looney folks out there doesn’t mean there isn’t an opportunity to make money. Gracious but firm negotiating can get you farther than simply giving up.
Of course it depends on the market value of the work.
Of course it depends on the market value of the work.
From my point of view, the open source and free software succeeded and went to it's apogee. But now, as noticed by this article, companies noticed that they could make a lot of money with that, and so there is a corporate take over of companies that don't have the core values but see the marketing interest of the oss.
So, little by little they used their money to get control of the main oss software/stacks and impose their flawed values. Look at npmjs, github bought by Microsoft, the Linux community that is now mostly corporate, os like Ubuntu, shitty tools like docker and mongodb and co...
So, little by little they used their money to get control of the main oss software/stacks and impose their flawed values. Look at npmjs, github bought by Microsoft, the Linux community that is now mostly corporate, os like Ubuntu, shitty tools like docker and mongodb and co...
Who ever assumed corporations were going to save the Linux community?
All the people who think VScode on Linux is great, and think it's nice that MS money backing github is not risky, and that WSL is cool and we should let them bring DirextX to Linux too.
In isolation any of those could be cool, but some see the totality of it as awesome rather than a full assault.
In isolation any of those could be cool, but some see the totality of it as awesome rather than a full assault.
Even worse, if you mention the "telemetry" in VSCode, you get a PR barrage of downvotes and comments saying it's "optional" and Microsoft isn't like facebook and they don't depend on data. Nevermind that data is the direction nadella wants to take microsoft - free windows 10, linkedin, github, etc.
The "love" you see for microsoft, gates, etc here is bizarre.
The "love" you see for microsoft, gates, etc here is bizarre.
> we should let them bring DirextX to Linux too.
The thing I like about open source is that you don’t need to ask permission to modify, expand, or target. Still, I would never run DirectX on my computers.
The thing I like about open source is that you don’t need to ask permission to modify, expand, or target. Still, I would never run DirectX on my computers.
Microsoft "Loves" Linux and open source the same way a tapeworm loves a healthy digestive system.
There seem to be some such people in this very discussion who think corporations might.
Red Hat is exceptional.
Pytorch and matplotlib managed to unseat matlab, which I didn’t think would ever happen.
Bitcoin, git are open source.
docker, kubernetes will just add a level of abstraction over any proprietary os.
Things are not so bad!
Bitcoin, git are open source.
docker, kubernetes will just add a level of abstraction over any proprietary os.
Things are not so bad!
"Pytorch and matplotlib managed to unseat matlab"
From what I know, most companies use MATLAB for real-time simulation or its VHDL code generation, which means you can put your algorithms into FPGA with relative ease. I don't see Pytorch helping there anytime soon.
That said, I agree with you that a lot of mathematical computing is moving to Python, especially now that there's multiple frameworks with GPU support. And yes, that's a great development :)
From what I know, most companies use MATLAB for real-time simulation or its VHDL code generation, which means you can put your algorithms into FPGA with relative ease. I don't see Pytorch helping there anytime soon.
That said, I agree with you that a lot of mathematical computing is moving to Python, especially now that there's multiple frameworks with GPU support. And yes, that's a great development :)
I came here to say something similar, but with a different industry. None of the people/groups I know using matlab have moved to pytorch, or even gnu octave which is a much closer fit.
Because its not really about matlab so much has the huge collection of library/extensions for matlab or its close relationship to simulink. Decades of industry specific modeling/simulation/tooling built into peoples workflows are hard to change.
If you look at the "applications" section here https://www.mathworks.com/products.html?s_tid=gn_ps pytorch really only touches on a couple of them AFAIK.
Deep engineering tooling is one of the places where opensource really hasn't yet made inroads. From the lack of a good solidworks/cad toolchain to the modeling/simulation space. Even EDA remains deeply dependent on proprietary tooling despite the recent success in a number of places. So while board level design with kicad may start eating into the Altium space now that its beginning to get RF/etc capabilities. I don't see it biting into the cadence/synopsis/mentor space anytime in the near future if ever due to the secrecy of the foundries.
I think a large part of that is there is a lot of breath in this space, and the customer base is generally quite small.
So I think a good opensource PCB layout/modeling tool, and a decent parametric 3d cad program will happen a lot sooner.
Because its not really about matlab so much has the huge collection of library/extensions for matlab or its close relationship to simulink. Decades of industry specific modeling/simulation/tooling built into peoples workflows are hard to change.
If you look at the "applications" section here https://www.mathworks.com/products.html?s_tid=gn_ps pytorch really only touches on a couple of them AFAIK.
Deep engineering tooling is one of the places where opensource really hasn't yet made inroads. From the lack of a good solidworks/cad toolchain to the modeling/simulation space. Even EDA remains deeply dependent on proprietary tooling despite the recent success in a number of places. So while board level design with kicad may start eating into the Altium space now that its beginning to get RF/etc capabilities. I don't see it biting into the cadence/synopsis/mentor space anytime in the near future if ever due to the secrecy of the foundries.
I think a large part of that is there is a lot of breath in this space, and the customer base is generally quite small.
So I think a good opensource PCB layout/modeling tool, and a decent parametric 3d cad program will happen a lot sooner.
Whoa, matlab has vhdl? I never knew, that's cool.
So, we need an open source license that makes the software not free for large corporations (say > $10M revenue).
Check out the Polyform Project's Small Business License. It has a threshold of 100 employees and contractors, and 1,000,000 USD in annual revenue.
Of course, the OSI's Open Source Definition precludes these kinds of restrictions, so it isn't "open source" by that definition.
https://polyformproject.org/licenses/small-business/1.0.0/
Of course, the OSI's Open Source Definition precludes these kinds of restrictions, so it isn't "open source" by that definition.
https://polyformproject.org/licenses/small-business/1.0.0/
> Of course, the OSI's Open Source Definition precludes these kinds of restrictions, so isn't "open source".
One would almost think there are some large corporations behind that organization and its definitions ...
One would almost think there are some large corporations behind that organization and its definitions ...
Actually these kinds of restrictions have always made the software "non-free" all the way back since GNU. Long before any large corps got involved.
OSS inherited this and was always very clear with it also.
OSS inherited this and was always very clear with it also.
I believe we need a way to charge companies of any size for the value that they generate out of using open source tools as part of their SaaS, PaaS, IaaS, etc. offering. If your cloud makes money by hosting an open source project as a service, that open source project should be entitled to a share of the revenue.
And, actually, we already have something quite similar, but I predict it won't fly with open source communities, most of which hate it unequivocally:
patents
And, actually, we already have something quite similar, but I predict it won't fly with open source communities, most of which hate it unequivocally:
patents
Patents are broken. But they can perhaps be fixed by effectuating a challenge-period after a patent application starts, whereby the patent can be invalidated if someone comes up with an implementation. Also there should be better protection against trolling, for example by requiring the holder of the patent to have a product implementing that patent. Further, a patent should not provide a strict monopoly on the technology for a given period (blocking competitors), but the rules should be so that anyone can buy a license for a reasonable price.
Dual license: AGPL and commercial. Simple.
AGPL requires you to disclose any changes to the source you make to an SaaS. But it allows making money.
If you want to disallow making money you have to use an even stricter license.
If you want to disallow making money you have to use an even stricter license.
No free license disallows making money. But the requirement for making source code available means people dont need to pay.
To make money you need to offer something more than just one program. Adding value outside the code itself is viable.
To make money you need to offer something more than just one program. Adding value outside the code itself is viable.
The issue with patents was that they were used for silly landgrabbing of obvious stuff and locking them up for a long time.
Now that things have matured a bit, I think 5-10 year long patents would be perfectly healthy.
Now that things have matured a bit, I think 5-10 year long patents would be perfectly healthy.
AGPLv3 isn't precisely that, but it may as well be since it scares corporations like Google shitless.
I disagree. The AGPL would still allow someone to rake in millions without sharing if they run an unmodified binary of an open source project.
https://medium.com/swlh/understanding-the-agpl-the-most-misu...
https://medium.com/swlh/understanding-the-agpl-the-most-misu...
While that is a problem it's a fairly harmless one to have, the company can't profit any more off of your work than anybody else can and you're not really losing anything by the company doing this. Contrast this to using the GPL where companies can effectively treat your code as proprietary and profit immensely off of it at the expense of everybody else who publicly contribute to your work but don't receive any kickbacks from big companies.
That's why I say it's not precisely the license requested. But if your aim is to scare off corporations, it works pretty well.
You can write any license you want and claim it's an open source license. It won't be OSI approved of course and, you almost certainly won't build a meaningful community around it but there's nothing keeping you from creating one.
Why won't you build any meaningful community ? If you only charge from big companies , why does it matter for some lone developer?
And as for big companies , as long as there's decent value and reasonable fees, why won't they contribute(to also help themselves) ?
And as for big companies , as long as there's decent value and reasonable fees, why won't they contribute(to also help themselves) ?
Many big companies (especially vendors) do contribute with people and otherwise--because, as you say, it does often help themselves--and most significant developers of major open source projects are doing it as part of their day job.
There's a long history of licenses in the PC freeware/shareware space of being only for personal use/educational use/etc. That usage-based restrictions did not make it into the FOSS world is an important reason that it's been so successful IMO.
There's a long history of licenses in the PC freeware/shareware space of being only for personal use/educational use/etc. That usage-based restrictions did not make it into the FOSS world is an important reason that it's been so successful IMO.
But what if I could pay members of the community using profit made from selling licenses to corporations?
I hate to be flippant, but that's a well-known business model called "contracting". It's possible, absolutely, but it requires a substantial investment of time and effort in business concerns on top of the actual software development. In practice, most leaders of contracting businesses don't have time to write code, and most contracting businesses end up deciding they need to release closed-source extensions.
How would e.g. Mozilla fit in that view?
As far as I know Mozilla doesn't sell any software licenses. They were almost entirely subsidized by Netscape from 1998 to 2003, then by Google from 2005 to 2014, and by a variety of search engines competing for Firefox default status since then. That kind of "cash cow" model gives you a lot more organizational freedom than software licensing.
Then you've re-invented being a software vendor that can't even pay it's developers reliably?
Why not reliably?
And why would you compare it to a regular software "vendor" when it produces free open software for 99.9% of the people, and non-free open software for 0.1% which are large corporations?
And why would you compare it to a regular software "vendor" when it produces free open software for 99.9% of the people, and non-free open software for 0.1% which are large corporations?
AGPL will prevent them from using it for SaaS. A dual license will then force them to pay creators.
If they are otherwise "selling" GPLed software that's kind of on the buyer to notice they dont need BigCorp to use it.
If they are otherwise "selling" GPLed software that's kind of on the buyer to notice they dont need BigCorp to use it.
So in essence a closed source license.
Part 6 of the osi definition is: "No Discrimination Against Fields of Endeavor"
Part 6 of the osi definition is: "No Discrimination Against Fields of Endeavor"
That's the OSI definition. Not everyone may agree on that definition especially as views change. Only one thing everyone agrees on is that "open source" is software for which the source code is published, and an "open source license" is a license attached to such software. This is not an extermely useful definition as far as the license is concerned, so if you want to be more specific you (and OSI) have to use more specific terms instead of the broad term "open source".
I disagree, otherwise we end up down in the rabbit hole of things like microsoft's "shared source" BS. Where terms loose meaning.
Its not just OSI either, its just most explicit there. This idea is also incompatible with DFSG and FSF definitions. It simply isn't FOSS to have license terms only apply to people you like. Either its Free for everyone or you are making glorified shareware/freeware.
Its not just OSI either, its just most explicit there. This idea is also incompatible with DFSG and FSF definitions. It simply isn't FOSS to have license terms only apply to people you like. Either its Free for everyone or you are making glorified shareware/freeware.
> WHEN the Linux Foundation (LF) let Microsoft in about 4 years ago we knew it was the beginning of the end
The good thing is that they knew. Now they hopefully will rush to regroup in a new setting before it is too late.
The good thing is that they knew. Now they hopefully will rush to regroup in a new setting before it is too late.
The GNU project makes it very clear that they are striving for the freedom of the user to modify and adapt the software that they use. So every time a GNU tool is used in the private back-end part of a SaaS webapp, that's directly opposite to the spirit of the GNU project, because you, the end user, have no way of modifying said SaaS.
And it's not only free speech in source code, it appears to me that free speech in general is under attack, now that people have noticed that a twitter storm and a supposed code of conduct violation is enough to wield immense power, for example by destroying someone's reputation, job, and life. We have effectively handed loaded guns to random strangers on the internet and, predictably, the results were bad.
The spirit of selfless sharing is also dying quickly, which I would blame onto the greed of internet startups and onto the entitlement of users that don't understand that GMail isn't truly free, just because there's no credit card charge.
I myself try to stay away from open source nowadays, because it ruins my mood if I get an entitled email from some random stranger who is insulting me because my free app did not solve his/her problem. In my case, a big company was kinda guilty in creating this problem, because they advocated my FOSS module on their homepage as if it was part of their expensive paid offering. So some of their customers feel like it's my duty to offer support...
Also, I would be very hesitant to release anything valuable with LGPL or even GPL these days. Chances are, it'll become the next cool "service" that Amazon, Google, and Microsoft will offer in their clouds, and I as the author of the software actually providing the service, will get nothing.
Case in point, one of my consulting clients is paying like $30,000 annually to Microsoft for hosted PostgreSQL. But apparently, re-selling PostgreSQL as a service with presumably millions in revenue is not enough for Microsoft to become an official sponsor... https://www.postgresql.org/about/sponsors/