One thing I've been thinking about lately - does the proliferation of web services harm free software?
In a way, no - a lot of web applications are actually based on free software.
However, a large part of human/computer interaction is moving from programs running on your local machine to programs running on an internet server.
And this means that those offering those services are not actually distributing or releasing software - in the case of CMS' and Web servers, they are only *using* it, and in the case of more specific web services, letting people use them as a *service* - in both case incurring no "redistribution" obligations under e.g. the GPL.
This means that even when more and more free software becomes available, people might still become increasingly obliged to use non-free software through the web, or to use free software as web applications under conditions where the four freedoms don't apply. I'm sure the question isn't new and others could state it much more eloquently, but ...
is the an answer, and which would be the right one?
best regards
Carsten Agger, Aarhus, Denmark
On Wed, 2007-07-18 at 17:10 +0200, Carsten Agger wrote:
This means that even when more and more free software becomes available, people might still become increasingly obliged to use non-free software through the web, or to use free software as web applications under conditions where the four freedoms don't apply. I'm sure the question isn't new and others could state it much more eloquently, but ...
is the an answer, and which would be the right one?
At the end of the day, people running software privately don't have to share it. Someone who modifies a GPL'd web app and makes the improved version available for use doesn't have to share their private modifications - they have the rights to share it, they just don't want to, much as I might improve my server's copy of Apache and not give copies to the people who use that server.
I don't think this is really about being free software or not free software; where someone declines to give out copies of something they have it's a privacy thing, not a freedom thing, and free software licenses respect privacy.
I suspect the real issue here is whether or not people have the ability to move from one service to another, and bring their data with them. That's an open standards argument, much like DRM.
You might be interested in some of the discussions going on around GUADEC, vis.:
http://tieguy.org/blog/2007/07/18/four-quick-notes-on-havocs-keynote/
Cheers,
Alex.
Alex Hudson wrote:
At the end of the day, people running software privately don't have to share it. Someone who modifies a GPL'd web app and makes the improved version available for use doesn't have to share their private modifications - they have the rights to share it, they just don't want to, much as I might improve my server's copy of Apache and not give copies to the people who use that server.
I don't think this is really about being free software or not free software; where someone declines to give out copies of something they have it's a privacy thing, not a freedom thing, and free software licenses respect privacy.
I suspect the real issue here is whether or not people have the ability to move from one service to another, and bring their data with them. That's an open standards argument, much like DRM.
You might be interested in some of the discussions going on around GUADEC, vis.:
http://tieguy.org/blog/2007/07/18/four-quick-notes-on-havocs-keynote/
I think the assumptions in that note are only partly true; yes, the technical side may give us an advantage, but the financial side (the ability to provide very large system with lots of bandwidth to support it) definitely plays into their hands. For example, we will not easily get an equivalent to google's massive datacentres.
And while this may not strictly be a free versus non-free issue, software services such as salesforce.com can easily make non-service equivalents irrelevant; at which point, by definition, free software also becomes irrelevant. The open standards argument can then become just that - an argument, with nothing to back it up (in the sense that 'there should be free operating systems' is not merely an argument, since it has been backed up by creating them).
Cheers Graham
On Wed, 2007-07-18 at 22:01 +0100, graham wrote:
I think the assumptions in that note are only partly true; yes, the technical side may give us an advantage, but the financial side (the ability to provide very large system with lots of bandwidth to support it) definitely plays into their hands. For example, we will not easily get an equivalent to google's massive datacentres.
As you note, that's not strictly a free vs. non-free thing, but I think there are a couple of things you haven't noticed.
First, free software is very good at being decentralised: git, tla/bzr, bittorrent, openid, etc., are all innovations realised first as free software. The data centre thing becomes less of a problem if you're not centralised.
Second, it's a matter of money, but not much money. Hosting is _really_ cheap, and you can play with the big boys without needing global dark fiber. For example, have you seen Amazon's Elastic Compute Cloud (EC2)? You effectively get root access to virtual machines running on fantastic hardware, and you can scale that up to silly numbers, and you pay 18 cents per gigabyte data transfer. Free software services could easily be setup on that platform for very little money.
It does come down to the fact that the web services play is exactly that: a services thing, and to compete in that market, you have to offer the service. If people want the service, it's going to be difficult to persuade them that they should buy a non-service version of the same thing.
At the end of the day, though, it's simply another business model. Cinemas co-exist with home cinemas, DVD sales co-exist with DVD rentals. You can rent or buy most things in life, and while some things make much more sense as a service (e.g., paying someone to deliver your mail, or to provide you with an internet link) I don't think software is necessarily in that category. It might be - but it might even be that services stop being popular in the future (getting your milk delivered used to be very popular, now it's much rarer).
Originally, all software was delivered as a service on a mainframe at basically no cost - maybe we're just reverting back to how things used to be :)
Cheers,
Alex.
[just a note]
On Wed, Jul 18, 2007 at 09:25:22PM +0100, Alex Hudson wrote: []
First, free software is very good at being decentralised: git,
What do mean `git'? GNU Interactive Tools or that one authored by Linus Torvalds? Latter isn't free software it's open source.
Please, maintain difference. ____
On Wed, Jul 18, 2007 at 11:50:09PM +0200, Oleg Verych wrote:
[just a note]
On Wed, Jul 18, 2007 at 09:25:22PM +0100, Alex Hudson wrote: []
First, free software is very good at being decentralised: git,
What do mean `git'? GNU Interactive Tools or that one authored by Linus Torvalds? Latter isn't free software it's open source.
Please, maintain difference.
What do you mean by that?
rms@roque:~$ rpm -qi git | grep License Size : 0 License: GPL
Rui
On Wed, 2007-07-18 at 23:50 +0200, Oleg Verych wrote:
On Wed, Jul 18, 2007 at 09:25:22PM +0100, Alex Hudson wrote:
First, free software is very good at being decentralised: git,
What do mean `git'? GNU Interactive Tools or that one authored by Linus Torvalds? Latter isn't free software it's open source.
I meant the tool authored by Linus Torvalds.
You can call it "open source" if you like; I don't use that term except to refer to a movement, and as this is software I will call it Free Software since that's what it is.
Cheers,
Alex.
Oleg Verych wrote:
[just a note]
And just a sidenote: pay attention to the mailing list you're sending this to. Most of the list subscribers are well aware about these issues and presumably have already oriented themselves.
On Wed, Jul 18, 2007 at 09:25:22PM +0100, Alex Hudson wrote:
First, free software is very good at being decentralised: git,
What do mean `git'? GNU Interactive Tools or that one authored by Linus Torvalds? Latter isn't free software it's open source.
It is free software because the author(s) chose a license that is a free software license. The views of the authors regarding the free software movement and its goals are irrelevant. However, namely these antagonistic views are a reason (valid or not) for some people, including myself, not to use their software when viable alternaives exist. This is also irrelevant to the issue -- git (the VCS) is free software regardless.
On Thu, Jul 19, 2007 at 12:59:11AM +0300, Yavor Doganov wrote:
Oleg Verych wrote:
[just a note]
And just a sidenote: pay attention to the mailing list you're sending this to. Most of the list subscribers are well aware about these issues and presumably have already oriented themselves.
Thank you for the clarification. Nevertheless somebody seems doesn't see the difference unfortunately...
On Wed, Jul 18, 2007 at 09:25:22PM +0100, Alex Hudson wrote:
First, free software is very good at being decentralised: git,
What do mean `git'? GNU Interactive Tools or that one authored by Linus Torvalds? Latter isn't free software it's open source.
It is free software because the author(s) chose a license that is a free software license. The views of the authors regarding the free software movement and its goals are irrelevant. However, namely these antagonistic views are a reason (valid or not) for some people, including myself, not to use their software when viable alternaives exist. This is also irrelevant to the issue -- git (the VCS) is free software regardless.
Whatever. Really. I've subscribed here due to closed archives. Now i see why i should leave.
| I don't call Linux "Free Software". I haven't called it that for close to | ten years! Because I think the term "Open Source" is a lot better. `--alpine.LFD.0.98.0706141153550.14121@woody.linux-foundation.org--
If you care to prove your talk, please see more: alpine.LFD.0.98.0706181344080.3593@... alpine.LFD.0.98.0706141402400.14121@...
Talking is most easiest part of all that thing...
Goodbye and sorry for my bad English. ____
Oleg Verych wrote:
Nevertheless somebody seems doesn't see the difference unfortunately...
Oh, if somebody == me (you should have stated that in plain text if that's the case; obfuscating is unnecessary -- it makes me wonder whether it's irony, fear from truth, honest assertion or just "diplomatic" expression); I certainly see the difference. That's one of the reasons why organizations like FSFE exist (Disclaimer: I am most certainly not entitled to speak on behalf of FSFE).
Now i see why i should leave.
If your interests do not coincide with computer users' freedom, perhaps you should leave lists like these as discussion may seem "boring" to you. But I never suggested anything like this and I don't think that my response was rude.
| I don't call Linux "Free Software". I haven't called it that for | close to ten years! Because I think the term "Open Source" is a | lot better.
How he calls "Linux" doesn't really matter. Linus Torvalds has declared multiple times that he's "apolitical". One doesn't have to align with his personal views to use the programs he has authored -- he doesn't do that when he uses GNU software either. So he wouldn't really mind if one calls Linux "Republican Software", "Orthodox Software" or "Voodoo Software" -- he is "apolitical", right? A vast amount of the community calls it "free software" because that's what it is (partly, as it contains non-free software as well).
Talking is most easiest part of all that thing...
I disagree a lot: there are many people who write useful stuff that is free software and these people have very diffrent opinions about the goals of the free software movement -- some of them oppose them fiercly. But nevertheless they write their free software, for their own reasons. However, only a few people talk about free software and how important is to preserve our freedom -- in my view, these people are doing a hard and a crucially important job.
On 19-Jul-2007, Oleg Verych wrote:
| I don't call Linux "Free Software". I haven't called it that for close to | ten years! Because I think the term "Open Source" is a lot better. `--alpine.LFD.0.98.0706141153550.14121@woody.linux-foundation.org--
Whether Linus chooses to refer to "free software" has no effect on whether it *is* free software. Its recipients all have the essential freedoms like any other free software, therefore it is free software.
Linus can call a horse a hamster if he chooses, it doesn't change the nature of the thing. Nor does it mean others should stop calling a horse a horse.
* Oleg Verych olecom@flower.upol.cz [2007-07-19 00:37:02 +0200]:
Really. I've subscribed here due to closed archives.
This list has a public archive, which is located here: http://mail.fsfeurope.org/pipermail/discussion/
Best wishes, Matthias
Talking is most easiest part of all that thing...
By talking, one spreads awareness about the moral issues of free software. One cannot do that just by sitting and hacking away.
Oleg Verych olecom@flower.upol.cz wrote:
What do mean `git'? GNU Interactive Tools or that one authored by Linus Torvalds? Latter isn't free software it's open source.
Please, maintain difference.
Please, explain what you think the difference is and why git is open source but not free software.
Puzzled,
First, free software is very good at being decentralised: git,
What do mean `git'? GNU Interactive Tools or that one authored by Linus Torvalds? Latter isn't free software it's open source.
git (the thing by Torvalds and co.) is indeed free software, it is licensed under the GNU General Public license; a free software license.
graham graham@theseamans.net wrote:
[...] For example, we will not easily get an equivalent to google's massive datacentres.
We already have an equivalent to it (our workstations) - it's just not very well-connected amongst itself.
I think open standards are useful even where free software does not exist, because it leaves a door open for us. We should ask the same things of these web services that we ask of other computer users who may or may not be using proprietary software. Send us open standard files not Word attachments, and so on.
Alex Hudson home@alexhudson.com wrote:
At the end of the day, people running software privately don't have to share it. Someone who modifies a GPL'd web app and makes the improved version available for use doesn't have to share their private modifications - they have the rights to share it, they just don't want to, much as I might improve my server's copy of Apache and not give copies to the people who use that server.
We also need to develop open service models which can be implemented in free software and are resistant against an attack of embrace-extend-extinguish. We have some of these, but we need more.
I feel that placing restrictions on the output of software, as seen in Affero GPL, is an evolutionary dead-end and a way for FSF to defeat itself. The complications of GPLv3 are already a disturbance (it may be more lawyer-friendly, but it's not nice to even copyright-experienced hackers, let alone hackers who don't understand copyright yet) and that may be a threat to free software. I hope that Affero GPL gets delayed, at least long enough to allow developers to digest GPLv3 and to let FSF webmasters unbreak stet, or maybe forever.
Regards,
On Thu, 2007-07-19 at 12:07 +0100, MJ Ray wrote:
We also need to develop open service models which can be implemented in free software and are resistant against an attack of embrace-extend-extinguish. We have some of these, but we need more.
Indeed, particularly, being able to run a distributed service with similar levels of reliability as a centralised service.
I feel that placing restrictions on the output of software, as seen in Affero GPL, is an evolutionary dead-end and a way for FSF to defeat itself. The complications of GPLv3 are already a disturbance (it may be more lawyer-friendly, but it's not nice to even copyright-experienced hackers, let alone hackers who don't understand copyright yet) and that may be a threat to free software. I hope that Affero GPL gets delayed, at least long enough to allow developers to digest GPLv3 and to let FSF webmasters unbreak stet, or maybe forever.
Seems to me the current Affero draft is mostly a no-op anyway:
"You may make, run and propagate covered works that you do not convey, without conditions so long as your license otherwise remains in force" (sec. 2)
=> if I don't convey and have a license, I have no obligations.
"Mere interaction with a user through a computer network, with no transfer of a copy, is not conveying" (sec. 0)
=> network interaction isn't conveying.
"Notwithstanding any other provision of this License..." (start of sec. 13, Affero draft)
=> the Affero clause doesn't withstand 2 & 0, and they say network interaction doesn't imply obligation, therefore it's null?
It will be interesting to see how that actually gets resolved. I'm not sure how the Affero could be a simple extension to the GPLv3 without somehow overriding the basic promises made in sections 0 and 2, and the obvious connotations of unfreeness that brings.
Cheers,
Alex.
I feel that placing restrictions on the output of software, as seen in Affero GPL, is an evolutionary dead-end and a way for FSF to defeat itself. The complications of GPLv3 are already a disturbance (it may be more lawyer-friendly, but it's not nice to even copyright-experienced hackers, let alone hackers who don't understand copyright yet) and that may be a threat to free software.
How is the GPLv3 less friendly to hackers? And how exactly is it more complicated than the GPLv2?
I hope that Affero GPL gets delayed, at least long enough to allow developers to digest GPLv3 and to let FSF webmasters unbreak stet, or maybe forever.
Without the GNU AGPL, we will have less means to protect our rights, so delaying it for a short or even forever would be a huge blow to free software. The original Affero GPL isn't compatible with the GPLv3 after all...
On Sat, 2007-07-21 at 14:52 +0200, Alfred M. Szmidt wrote:
How is the GPLv3 less friendly to hackers? And how exactly is it more complicated than the GPLv2?
It's longer and invents new language - it seems pretty factual to me that it's a more complex license, and the GPLv2 wasn't the most easily understood license in the first place (hence so many people making rudimentary mistakes like "You can't sell GPL'd software").
That's not saying it does that unnecessarily, but it does seem to be indisputable.
Without the GNU AGPL, we will have less means to protect our rights, so delaying it for a short or even forever would be a huge blow to free software.
Not everyone agrees that the right to see software source on someone else's machine you're using is a free software right; I'm not particularly sure I do.
The original Affero GPL isn't compatible with the GPLv3 after all...
That's a shame if it's not, they did build in a clause to make it compatible:
"You may also choose to redistribute modified versions of this program under any version of the Free Software Foundation's GNU General Public License version 3 or higher, so long as that version of the GNU GPL includes terms and conditions substantially equivalent to those of this license."
Perhaps that upgrade route is dead now the Affero clause didn't make it into the GNU GPLv3.
Cheers,
Alex.
How is the GPLv3 less friendly to hackers? And how exactly is it more complicated than the GPLv2?
It's longer and invents new language - it seems pretty factual to me that it's a more complex license, and the GPLv2 wasn't the most easily understood license in the first place (hence so many people making rudimentary mistakes like "You can't sell GPL'd software").
While the GPL (any version) is not a trivial license, any hacker who is capable of writting a non-trivial program should be able to grasp it in an hour. As for terms, hackers often invent terms to be able to be convey a more precise meaning. Still, it is a easy license compared to most other licenses, and the general ideas are easily grasped by the four freedoms of free software.
Without the GNU AGPL, we will have less means to protect our rights, so delaying it for a short or even forever would be a huge blow to free software.
Not everyone agrees that the right to see software source on someone else's machine you're using is a free software right; I'm not particularly sure I do.
I think that this is no different than a machine that I own that prohibits me from upgrading it.
The original Affero GPL isn't compatible with the GPLv3 after all...
That's a shame if it's not, they did build in a clause to make it compatible:
"You may also choose to redistribute modified versions of this program under any version of the Free Software Foundation's GNU General Public License version 3 or higher, so long as that version of the GNU GPL includes terms and conditions substantially equivalent to those of this license."
Perhaps that upgrade route is dead now the Affero clause didn't make it into the GNU GPLv3.
This clause has nothing to do with the Affero license though. The Affero license isn't "substanitally equivalent" to GPLv3. This is why the GPLv3 contains the following text:
| 13. Use with the GNU Affero General Public License. | | Notwithstanding any other provision of this License, you have | permission to link or combine any covered work with a work licensed | under version 3 of the GNU Affero General Public License into a single | combined work, and to convey the resulting work. The terms of this | License will continue to apply to the part which is the covered work, | but the special requirements of the GNU Affero General Public License, | section 13, concerning interaction through a network will apply to the | combination as such.
When I spoke about the Affero GPL I meant the original one, and not the GNU Affero GPL which is yet to be published.
On Sun, 2007-07-22 at 13:20 +0200, Alfred M. Szmidt wrote:
While the GPL (any version) is not a trivial license, any hacker who is capable of writting a non-trivial program should be able to grasp it in an hour.
Perhaps "should", but they don't.
Still, it is a easy license compared to most other licenses, and the general ideas are easily grasped by the four freedoms of free software.
Compared to most other licences? I'm not sure about that. It's more complex than most BSD-like and Apache-style licences, which are a significant proportion of "other".
Not everyone agrees that the right to see software source on someone else's machine you're using is a free software right; I'm not particularly sure I do.
I think that this is no different than a machine that I own that prohibits me from upgrading it.
Whereas I think it's no different to using a shell on a shared server.
That's a shame if it's not, they did build in a clause to make it compatible:
"You may also choose to redistribute modified versions of this program under any version of the Free Software Foundation's GNU General Public License version 3 or higher, so long as that version of the GNU GPL includes terms and conditions substantially equivalent to those of this license."Perhaps that upgrade route is dead now the Affero clause didn't make it into the GNU GPLv3.
This clause has nothing to do with the Affero license though.
Er, yes it does - it's a clause in the original Affero licence, which was supposed to provide GPLv3 upgrade capability similar to the LGPL.
The Affero license isn't "substanitally equivalent" to GPLv3.
Indeed, so it looks like the Affero -> GPLv3 upgrade route is dead, since the original Affero doesn't mention the possibility of relicensing to the GNU Affero GPL.
Cheers,
alex.
While the GPL (any version) is not a trivial license, any hacker who is capable of writting a non-trivial program should be able to grasp it in an hour.
Perhaps "should", but they don't.
Probobly for the same reason they don't write well written programs... ;-)
Still, it is a easy license compared to most other licenses, and the general ideas are easily grasped by the four freedoms of free software.
Compared to most other licences? I'm not sure about that. It's more complex than most BSD-like and Apache-style licences, which are a significant proportion of "other".
I suppose that is what I get for being vauge, by most other licenses, I was refering to non-free software licenses.
Not everyone agrees that the right to see software source on someone else's machine you're using is a free software right; I'm not particularly sure I do.
I think that this is no different than a machine that I own that prohibits me from upgrading it.
Whereas I think it's no different to using a shell on a shared server.
Well, then I can only assume that you think what Tivio did is ok? Since that is one way to look at it, a hardware with a shell that the company is providing access to.
That's a shame if it's not, they did build in a clause to make it compatible:
"You may also choose to redistribute modified versions of this program under any version of the Free Software Foundation's GNU General Public License version 3 or higher, so long as that version of the GNU GPL includes terms and conditions substantially equivalent to those of this license."
Scratch what I wrote after this, I thought this was from the GPLv3...
On Sun, 2007-07-22 at 18:45 +0200, Alfred M. Szmidt wrote:
Compared to most other licences? I'm not sure about that. [GPLv3]'s more complex than most BSD-like and Apache-style licences, which are a significant proportion of "other".
I suppose that is what I get for being vauge, by most other licenses, I was refering to non-free software licenses.
Ah, ok. I suspect MJ's point was related to other free software licences, but he can clarify that if not.
Well, then I can only assume that you think what Tivio did is ok? Since that is one way to look at it, a hardware with a shell that the company is providing access to.
No, I think they're objectively different situations. With Tivo, they were selling hardware that had modified free software on them: the user, having purchased the hardware, was unable to change the software on it, but Tivo could.
With a shared shell, I'm not buying the hardware - I've bought a service. That's very different - it's like buying web hosting, or something like that. Personally, if I buy web hosting service, I don't expect the host to offer me the source to their Apache, for example.
So, primarily, I don't see that web applications should get special treatment, and I wouldn't like to see such a feature in (e.g) CLI applications. But second, I just don't see how it's a free software issue. If I don't give someone a copy of a piece of software, then I shouldn't have an obligation to them, and I don't see why people should be forced to give out copies of software. It's completely different to the Tivo situation, where they actually _are_ distributing the software.
Cheers,
Alex.
Alex Hudson home@alexhudson.com wrote:
Ah, ok. I suspect MJ's point was related to other free software licences, but he can clarify that if not.
I actually meant GPLv3 was less hacker-friendly than GPLv2, but GPLv2 was already less hacker-friendly than many other free software licences. Just the sheer length and complexity of concepts made it that way. In GPLv2, much of that seemed necessary. With GPLv3, I'm not sure yet.
[...]
So, primarily, I don't see that web applications should get special treatment, and I wouldn't like to see such a feature in (e.g) CLI applications. [...]
I don't see how requiring programmers to include more than a simple information statement in a program's output is acceptable for free software. IIRC, AGPL seemed to claim it was to do with that output being a network-based user interface, but that's a very dodgy line crossed from the GPL, which doesn't place obnoxious restrictions on program output. For example, do survey form design programs have a paper-based user interface or is that merely output? Why should the freedom standard differ if that form design is to HTTP not paper?
Regards,
"Carsten Agger" agger@c.dk writes:
This means that even when more and more free software becomes available, people might still become increasingly obliged to use non-free software through the web, or to use free software as web applications under conditions where the four freedoms don't apply. I'm sure the question isn't new and others could state it much more eloquently, but ...
is the an answer, and which would be the right one?
The question isn't that new. There even exists a license for this question, the Affero GPL[0]. The FSF currently works the GNU Affero GPLv3[1] which will be compatible to GPLv3.
Simplified the Affero GPL says that if you use Free Software licensed under the Affero GPL as a web service you have to give all users of this service a copy of the source code and all rights defined by the Affero GPL. Basically it extends "distribute" to "offering web services".
I think the crucial point is that if you use such a web service ultimately you don't know what software runs on the server and what will happen to your data. So whether you will get a copy of the source code or not, you become dependent on the service provider and lose control over your computing.
Personally i would avoid using web services whenever it is feasible and prefer to use software that runs on my own computer so that i have the full control over it.
[0] http://www.affero.org/oagpl.html [1] http://gplv3.fsf.org/agplv3-dd1-guide.html