Hacker News Reader

Comment by pydry | Hacker News Reader

That's another license that gets a disproportionate level of flak.

Big companies or those with lawyers notice that the AGPLv3 says the software used to make something a service must be licensed under the same license. Much of that software isn't something a company controls. For example, you put the AGPLv3 software behind a load balancer. Does that software need to be AGPLv3? Some would say so and how often can a company control that license.

Companies with more are conservative on their risk.

Not a lawyer and this is not legal advice

>Does that software need to be AGPLv3?

No. It doesn't "infect" the load balancer. There's zero ambiguity about this.

A lot of companies have bad lawyers who try to eliminate rather than mitigate risk and who try to grab every scrap of IP for the company. This is the kind of company you're talking about.

This is also why many try to get you to sign away every profitable idea you've ever dreamed up in the shower: a combination of extreme risk aversion and flagrant greed.

I'd even venture as far as to say it's a feature not a bug if this type of company were forced to use expensive proprietary software or older, shittier versions.

> No. It doesn't "infect" the load balancer. There's zero ambiguity about this.

This is not true.

> A lot of companies have bad lawyers who try to eliminate rather than mitigate risk and who try to grab every scrap of IP for the company. This is the kind of company you're talking about.

I'm guessing you would claim all of the giants as obviously falling into this category (they all have bad lawyers).

They are all also the only ones worth suing.

Note that it's not just FAAMNG, Square made this determination too: https://hub.packtpub.com/square-updated-its-terms-of-service...

I'd love to hear what Stripe thinks about using AGPL-licensed software.

Basically I've yet to hear of a large (> $1B revenue +) SaaS company that doesn't have concerns about using AGPL software as a core part of its service offering. I'd love to hear some counterexamples...

>This is not true.

Please stop spreading misinformation.

>I'm guessing you would claim all of the giants as obviously falling into this category

Google definitely counts as a company that tries to grab every scrap of IP for itself. They also heavily customize everything and compete in all software markets. For them this is ideological because it interferes with their profits as a monopoly. They want the license dead and buried because it targets them.

This is a feature, not a bug.

Small Podunk companies that shy away from installing mongo, OTOH are being legitimately dumb.

Yeah. The boundaries of the GPL are mostly clear because they're essentially tied to Unix linking behavior. There are some edge case--I know people who don't think you shouldn't be able to load binary blobs for example--but mostly. There doesn't seem to be the same sort of clear consensus over how broadly the AGPL interacts with other code on the network.

AGPLv3 get the exact same level of flak as GPL got. And will lose (lost?) in exactly the same way.

GPL was the fight against device manufacturers using linux et al and not giving back/selling closed source linux devices.

We completely lost with tainted kernel and such, as they corrupted the only software that they couldn't live without and was promoting an open source license. Case in point: you cannot build 1% of your android phone software, proving GPL code (android, linux kernel) is as useless for open software as MIT (ios, darwin)

Now the fight moved from OEM manufacturers vs GPL to cloud providers vs AGPLv3. And source-available licenses are the tainted-kernel compromise all over again: get the thing you cannot live without but is fighting you with a pro-opensource license, and offer a carrot so they change, and they all changed.

Without the change the article talks about, cloud providers would have people using the AGPLv3 code in their derivative work of projectX. while they paid the closed source version from the company dual licensing it. With this new arrangement, they can use all derivative work at will with zero consequence, for the same low price.

Just like tainted kernel was a hard blow in the face of everyone who contributed to linux (heh, specially the GNU folks porting their stuff) with open source in mind. This is nothing but a greedy bait and switch on the community.

I often see not 'just' AGPL, but AGPL+Source Attribution Policy (if you contribute to the main repo, you must assign copyright to the owner), and specifically around network/web oriented software.

This combination is pretty antithetical to the 'spirit' of open source (and especially 'libre software', which AGPL is supposed to be) since it basically ensures hegemony on any commercialization ecosystem to the original developer, unless one manages to create a 'leading fork' where contributions land, since only the original rights owner (who also gets the rights to the sum of all other contributions) has the legal means to commercialize (i.e: provide commercial hosting around) any extensions above and beyond what the community develops.

>This combination is pretty antithetical to the 'spirit' of open source (and especially 'libre software', which AGPL is supposed to be) since it basically ensures hegemony on any commercialization ecosystem to the original developer

The spirit of open source was never about giving away the software for free.

https://www.gnu.org/philosophy/selling.en.html

The community was concerned with free speech. You want free beer (freedom of commercialization).

AGPL also says that you have to make your changes available. It doesn't say you have to sign a CLA. If you value your "IP", don't. If you value having your changes pushed upstream, do. It's completely up to you.

In terms of 'freeing the source' AGPL without source attribution is just as 'libre' as with it - but with it, the effects I mention are possible - therefore it's radically different when used this way than with the license alone, and so has a 'different spirit'.

You are right though, that 'libre' software is not about economic freedom but keeping the software 'free'. One of the reasons I'm in the BSD/MIT camp - software doesn't exist in a vacuum.

Reply on news.ycombinator.com