Hacker News Reader

Comment by autarch | Hacker News Reader

What do you hope to achieve with this? Ok, you win, the term "open source" predates the OSI. So what?

Using the term "open source" without any definition is useless. If we can't agree on a definition, it's impossible to know if we're actually talking about the same thing.

I want people to use the OSI definition in order to elevate debates. I'd prefer to skip past definitions to more substantial matters, like whether "open source" (per OSI) is useful. Is it somehow better than closed source code? Is it _ethically_ valuable? Is there some subset of the OSI definition that provides more value than the rest? These are interesting discussions worth having.

Endless debating the meaning of "open source" is a huge waste of time.

Given that OSI is the only body I know of with a clear definition, let's use theirs and move on to more substantial topics.

View on HN

> Given that OSI is the only body I know of with a clear definition

No, you can also use the common definition of "open source" = "not closed-source" = "not (source unavailable)". Nobody has branded this definition but that doesn't make it any less legitimate. See definition #1 on dictionary.com for "closed-source", or #2 for "open-source". [1] [2]

> I want people to use the OSI definition in order to elevate debates.

This is... obviously biased? Other people prefer to use other definitions to elevate debates. You can't claim only the definition you like is able to elevate debates.

And the parent is putting so much effort into arguing about the definition for the same reason you did in your comment. If it was so inconsequential, nobody would care. But evidently people find it a powerful thing, hence they argue about it. You can't simultaneously do that and then claim it's irrelevant.

[1] https://www.dictionary.com/browse/closed-source

[2] https://www.dictionary.com/browse/open-source

I think this is the most sensible and inclusive definition, otherwise you have a lot of situations where it's not technically OSI "Open Source" but the source is literally open.

I've seen people use "source available" (?) in these situations, but I don't think it really makes sense because a lot of the time the only thing holding it back from being OSI "Open Source" is that their license has not been recognized by OSI.

But now we need a new term to mean what "open-source" has meant for two decades, just because for some reason we wanted to be inclusive of licenses where the source is viewable but not open for use. And once we've redefined it, we've rendered all discussion of open-source deceptive for the period where it had its traditional meaning. I don't see any benefit to this inclusion.

I would like a term that is inclusive of CC0.

Creative Commons licenses aren’t recommend for software[1], but you can use the Zero-Clause BSD License[2] to satisfy the same goals as CC0.

[1] https://creativecommons.org/faq/#can-i-apply-a-creative-comm...

[2] https://opensource.org/licenses/0BSD

from [1]

> Also, the CC0 Public Domain Dedication is GPL-compatible and acceptable for software. For details, see the relevant CC0 FAQ entry.

Ah didn’t realize that. Thanks for pointing it out.

It's applicable to software, but it's not a good software license due to reservation of patent rights. CC was designed for scientific publications, in that context reservation of patent rights has an understandable meaning, but when applied to software it takes a very different unintended meaning.

How about "copyfree"?

https://copyfree.org

https://copyfree.org/standard

(I also consider anything that fits the Open Source Definition to be "open source" and differentiate things actually approved by the OSI as "OSI certified", which should in principle be a subset of "open source".)

> just because for some reason we wanted to be inclusive of licenses where the source is viewable but not open for use

This is not true. Recent licenses trying to protect the business built on the open source code are in general, open for use:

- Sentry: https://news.ycombinator.com/item?id=21466967

- Elastic: https://news.ycombinator.com/item?id=25833781

I see these sorts of licenses becoming increasingly common in the future, which is why I think it's silly to continue excluding them from being called open source.

The primary distinction in those licenses is that they're not generally open for use — they allow a carefully chosen, closed set of use cases. As an analogy, when a bar has a TV showing some preselected channel at a preselected volume, I don't consider that TV open to my use, even though I can use it for the use case the bar specifically chose to enable.

I do agree that licenses like this will become more common in the future, and that's why I think it's useful to have an identifying term for them rather than making "open-source" less precise to include them. Different words for different things is good, in my opinion.

> they allow a carefully chosen, closed set of use cases

I would argue that they prohibit far less use cases than they are open for.

In any case, how would you describe these licenses? I don't feel like "source available" is an accurate descriptor in this case.

I agree. My personal term for this sort of "We're OK with little people using the software but we don't want any competition" arrangement is "private-use source license," but I wouldn't be so bold as to argue that's The Best Name. My point is just that I don't think broadening "open-source" is a good answer, because all that does is make it harder to talk about the differences in licenses.

Disclosure: I work at Sentry.

> My personal term for this sort of "We're OK with little people using the software but we don't want any competition"

Large companies are free to use Sentry. There are Fortune 50 companies running Sentry at scale internally without paying us a cent. That's totally cool.

You're also free to compete with Sentry. You're not free to repackage Sentry for the purposes of competing with us. There are lots of competing error and performance monitoring products out there that do perfectly fine without it.

I should also note that many components of Sentry are distributed with OSI-approved licenses that you are free to use to compete with us. For example, our Symbolication service (https://github.com/getsentry/symbolicator) ships with an MIT license, and it's an important part of our business.

What's sentry's incentive to not relicense in the future so those fortune 50 companies have to pay you instead?

That seems like an obvious growth opportunity when investors need to see numbers go up

The incentive is that running the infrastructure required to host Sentry is challenging and requires engineering resources. You can choose to use our hosted services for a fraction of the cost (engineers are expensive).

Aside, there seems to be confusion about what “relicensing” means. Even in this hypothetical scenario you’ve outlined, we can’t “relicense” already released software. Many users hosting Sentry internally are using years-old versions happily; they would continue to be able to do so. They could also choose to fork the last permissible version and maintain it themselves.

Personally, I think this scenario (relicensing such that self-hosted users could no longer do so) is incredibly unlikely. I don’t believe it would really do anything to grow the company.

I would describe it as "Restricted use" software which has existed for decades, generally as closed source.

Sure the SaaS server licenses are more or less open use with caveat, but why won't a process paralleling tivoization occur?

I wouldn't find it odd for a hardware vendor to release source code, "for review" without the rights to use it on any other hardware as some do for binaries.

In a way that could be worse than closed source as that is similar to the problem of suspicion of reading leaked source code, maybe you are tainted and can't actually write software licensable for competing hardware. Maybe you never read the code yet it will appear that way, but you shouldn't check.

Given the oddities of IP and copyright infringement, it is often the case that only the owner benefits from a process of publishing. I.e. what good is the disclosure of a modern patent?

Yup, this is a huge thing. Get students to use the code, get comfortable with it, then when they write something for their local city or whatever, now that local city is potentially in violation.

It’s kind of like that thing where software vendors give cheap or free licenses to educational users in order to get them hooked and then charge whatever company they might work for an arm and a leg to keep using it.

>Different words for different things is good, in my opinion.

And yet people are fighting tooth and nail to redefine open source to mean OSI approved licence instead of using OSI approved license to mean OSI approved license.

>To see why just change the thread title: What’s up with these new non-OSI approved licenses?

It's an article that you can have a reasonable debate about without stacking the deck against the people you're arguing against. Which is why the people who make money by stacking the deck are so dead set to redefine the word to mean exactly what makes them the most money.

"Open source" and "free software" have always been mostly interchangeable and have since their inception not simply meant "source available".

The people trying to redefine anything are those ignorant of the history here.

>"Open source" and "free software" have always been mostly interchangeable and have since their inception not simply meant "source available".

Free software was coined in the 80s. I wasn't there for that.

Open source was coined in the 90s because we didn't want people to think that software which let you see the code was free of charge.

Then when they sold out a bunch of tech bros from SV decided that they knew best and redefined open source to mean whatever made their valuations go to the moon. The plan did work as expected because dot com crash.

The people around the OSI have always been share croppers that have been trying to get paid for other peoples work.

Writing "tech bro" and "sharecropper" makes it easy for people to stop trying to look for meaning in your writing.

Also, while I'm sure it's unintentional, your backwards of "sharecropper" smells racist.(It actually means paying someone for the right to do you your own work, and was an attempted to recreate slavery after slavery was banned in USA).

> redefined open source to mean whatever made their valuations go to the moon

Was it redefined? My impression is that it has always effectively been a synonym for free software, just not carrying the political leanings.

This is correct. 'Open Source' was a way to make it more palatable to business, but the definition was pretty much identical.

I think you are right "OSI approved license" is a much more precise term than "Open Source".

If OSI wanted to provide the official definition for what is "open source" and what is not, they could have perhaps trade-marked their term "Open Source".

But "OSI approved license" has its own vagueness about it. What does it mean that OSI "approves" a license? Approves it for what? To be called "Open Source"? But then unless they have the trademark for the term Open Source, they shouldn't be the one to give approvals for people to call their software that.

This reminds me of the debates about whether something is true Agile, or merely wannabe-Agile. Words are just words. It is good to have precise definitions but those can be specific to a given context. It is a bit like let x = y + z. Is that the correct definition of x? No, but it is a VALID definition of it, assuming y and z are well-defined as well.

They do have the trademark for the term Open Source.

Nope, only for "Open Source Initiative" and "Open Source Initiative Approved License", according to https://writing.kemitchell.com/2020/05/11/Open-Source-Proper...

> licenses trying to protect the business built on the open source code

A fundamental aspect of open source is that there is no limit to the number of businesses that can be built on it. If a single company wants a monopoly on building a business on the software, we have a term for that: proprietary software.

This exactly, people are crying "big tech, boo hoo", but the fact is these technology companies like elastic search and co thought they could make open source work for their business model but they struggled in the face of competition and so got defensive.

And this is the fundamental issue, can they build a strong company based on this software, it's potentially possible but that seem to think they can't be competitive hence the license moves.

They made a product in which they spent resources and time and provided as open source, then another company which has many more resources took their open sourced code (which legally they could), did some work (or not) on it and provided exactly the same service, behind their now closed source branch. Not only did this company have much more resources to allocate to it they didn't need to bear the discovery and development costs and took an already proved idea.

It is legal, it was a good business move by amazon but it definitively sucked for those who had invested in building that software.

So yeah, in a way I can understand that now anyone risking their business in creating software or services based on those might be going, "yeah no thanks, you can't fork our work and sell it".

Specially when companies are gigantic and a 0.01% of their budget equates to the whole budget of the company who produced the original source. You can't really "not struggle" in the face of the competition.

I have to dig more into GPL licensing but anyway, there certainly needs to be some licenses that are not "f$#%% me any way you want" for people doing open source products.

The license these businesses should be using is AGPL, clean and simple.

Run the software under AGPL and dual license it under a closed or restricted usage license if a customer doesn't want to use the AGPL license.

Instead we have people electing to create restrictive licenses with varying degrees of "use the software however you like unless we don't like you or your business" baked into them.

The dual full free + restrictive license model has worked very well in the past for FOSS based companies. Let people who are fine contributing back to your project and/or open their project up use it however they please and let everyone who doesn't like those terms pay for it like standard commercial software instead.

Can you think of any successful AGPL + commercial license examples? I like the idea

Off a quick google search Ghostscript, Bitwarden, Berkeley DB, and Instructure Canvas.

I like the idea, will have to read AGPL. In my mind an ideal version would be, when translated from legalese:

If your business is just an extension of the underlying software (the source) and you don't want to open source your part of the work, then pay a commercial license.

If your business uses the software (the source) just as an ancillary part most probably you're ok to source back any improvements back and keep them open source (no need to pay). If it's ancillary and you need to keep them closed, then pay for licensing anyway.

The only thing I don't understand but seen many time mentioned is the "infecting" nature of the GPL licensing?

The "infection" is the feature that makes GPL and AGPL useful. The basic idea is that if you are interfacing with the (A)GPL licensed software you are extending it's functionality and must comply with the license. The difference is that the interfacing with GPL is statically linking against the project. With AGPL the interfacing is also interfacing over the network. What this means is that software that clients or peers interacting with the software need to be available to the user. An additional stipulation for AGPL is that all necessary infrastructure to host the service be open sourced as well.

The reason for this is to prevent people from creating wrappers that allow them to profit off the licensed software without contributing back or open sourcing. The key thing to note is that you don't have to upstream changes. You only have to provide the source to users (and the license protections extend to them). This means you can still sell GPL and AGPL software but you do have to provide full usage permissions to the user once they've bought the product.

As a side note: I've always found it amusing that government contractors have traditionally been very afraid of (A)GPL despite being required to provide (A)GPL-esque usage permissions to the customer/government. Most of this I imagine is due to a fundamental misunderstanding that providing source to the user doesn't mean publicly make source available for everyone.

Now if you have an AGPL or GPL licensed dependency and you want to dual(+commercial) license your project, you have to work out a commercial license agreement with the dependency's maintainer. If a project has (A)GPL dependencies and doesn't have a commercial license but you'd like to use one, you'd have to work out commercial licenses with the project and all the dependencies (or have the project maintainer work said license out).

This infection effectively enforces users to either share their software or "pay" for all of it which is arguably a good thing for the health of the ecosystem as a whole. If (A)GPL+commercial licensure was standard practice, we likely wouldn't be seeing the [OpenSSL/xkcd-2347](https://xkcd.com/2347/) issue keep popping up time and again.

TLDR: The "infection" forces users/developers to either respect the "free" terms or pay for the full value of the software.

They tried to play a game with open source, they knew the rules and AWS did nothing wrong, not even ethically, it's all in the spirit and the letter of the licenses ES chose (for example).

I've literally no sympathy for ES whatsoever, they were either incompetent or playing games.

What ES invested and the size of AWS has literally nothing to do with the discussion, ES made a choice pure and simple. They are within their rights to make a new choice if they so choose.

> not even ethically

Ethics don't come into play here and I also don't have any specific sympathy for ES, but I have sympathy for people and companies that provide open source software that others can read and tinker with.

I can like AWS because of providing me services I find useful. That doesn't mean that I have to like everything about them.

> What ES invested and the size of AWS has literally nothing to do with the discussion

Well, it does and doesn't. In a way it would be great to have open-source software be the default, but given that anybody with resources can simply take it and build a then non-open-source version of it and polish it or integrate into much higher leverage existing systems, it does pose some questions as towards what licensing open-source software should have in the first place and if it should be infecting or simply have multiple different clauses related to how it's being used.

But yeah, when playing a game make sure you know the rules and all.

Imo ES got benefit from open source sa marketing tool. Amd no matter what AWS contributed back the fact is ES doesn't want someone else to reap the benefit of that marketing play.

>otherwise you have a lot of situations where it's not technically OSI "Open Source" but the source is literally open.

Just because the source is available to you doesn't mean it's open source though. There are many companies that have the source code to Windows, but I would never describe Windows as being open source. That where the pedanticism comes in.

Definition #1 is, while imprecise, more consistent with OSI open source than “source available”; “intellectual property” is a set of exclusive rights, not the content to which the rights apply—“source available” provides the content to the public, but not necessarily any substantial subset (or even necessarily any) of the rights constituting intellectual property.

> Other people prefer to use other definitions to elevate debates.

No, they'd prefer other definitions to obfuscate differences, specifically, when trying to define “open source” as equivalent to “sourve available” to erase the distinction that having separate terms with distinct definitions provides.

> What do you hope to achieve with this? Ok, you win, the term "open source" predates the OSI. So what?

The point is to demonstrate that the term predates the OSI's alleged coinage thereof. They don't get to dictate language. This usage pre-exists them and obviously persists to this day.

> Using the term "open source" without any definition is useless. If we can't agree on a definition, it's impossible to know if we're actually talking about the same thing.

Really? It seemed quite useful to the people cited in each of those earlier prior references. They seemed to know exactly what they were talking about in context. You have only to read the messages to see that.

> I want people to use the OSI definition in order to elevate debates.

No, I think you want people to use the OSI definition because that conveniently includes certain clauses that have nothing to do with being open source, much like the FSF's definition of free software has nothing to do with freedom.

> Endless debating the meaning of "open source" is a huge waste of time.

You're right. Equally useless is attempting to privilege the OSI's definition over others.

> Given that OSI is the only body I know of with a clear definition, let's use theirs and move on to more substantial topics.

The numerous citations you are now aware of make you aware of others with a clear definition, so... No.

Equally useless is attempting to privilege the OSI's definition over others.

The OSI definition was elevated over others (to the extent that there even are any others) by usage. You can dislike that all you want, but it doesn't change anything. Maybe, in time, usage will flip the meaning to something else... English has a way of doing that. But let's not stick our heads in the sand and pretend that current reality is anything other than what it is.

Language is not a popularity contest. Words and phrases can mean different things in different contexts. That the OSI came along and proffered its own definition does not mean we have to forget what came before and, moreover, is still current today. The OSI is an organization that co-opted a movement for their own business purposes... much like Eric Raymond co-opted the MIT Jargon File.

And companies releasing “open source” software which has hooks in it that put other companies in potential legal trouble (beyond their he usual GPL type stuff) are themselves co-opting the term for their own ends. It’s hiding behind “language flexibility” to use “open source” in this manner.

...or they're using that phrase exactly as it was intended before the OSI got their hands on it.

I haven't seen evidence of that.

There's links to plenty of it over here[1] which I posted up-chat.

[1] https://www.arp242.net/open-source.html

Your link doesn’t support your claim. Most of those seem closer to the OSI definition (BSD is the first one, GPL, Linux, etc) and the ones that might contradict it are rarer.

The way most people have been using “open source” is a lot closer to OSI than these companies are claiming. It’s misleading to use a different definition nowadays. And I think that’s the actual intent. If it wasn’t, then they’d be perfectly fine with using another unambiguous term like “source available.” If there was a strong desire to be clear and not misleading, they’d use such a term.

Language is not a popularity contest. Words and phrases can mean different things in different contexts.

Except.. it kinda is. I mean yes, it is obviously true that words can (and do) have multiple meanings. But within a given context, there are "generally accepted" definitions and there are idiosyncratic / obscure / unusual definitions. And in the kind of context we're talking about here, to use "Open Source' to mean anything other than what is defined by the OSD is to intentionally choose a non-standard, idiosyncratic definition. It wouldn't be reasonable for one to do that and then expect everybody else to understand / agree with their non-standard usage.

Naa. I was around when OSI was created. They weren't authoritative then, they're not authoritative now.

"They" don't have to dictate anything. The point is that "Opensource" as currently used means something else than shared source. Freeware means something else than free software and free speech does not mean zero cost cellphones.

Taken literally these words could mean something else, but they don't. And using them in a wrong, hoping that people will confuse them for something they like (free speech for phones is good, right?) is extremely dishonest.

> Using the term "open source" without any definition is useless. If we can't agree on a definition, it's impossible to know if we're actually talking about the same thing.

Which is why you should call them "OSI Licenses" if you are referring to OSI licenses.

> let's use theirs and move on to more substantial topics.

This just isn't how English works. It isn't prescriptive. English is descriptive by nature and permits multiple uses, which have been around before OSI ever existed and are still valid.

Sure, but I think companies are actually abusing the common definition of open source (which roughly has been the OSI one) by calling their source-available (but private licensed) stuff “open source.”

> Which is why you should call them "OSI Licenses" if you are referring to OSI licenses.

1. They aren't written by or copyright or anything to the OSI. They aren't "OSI Licenses" they are "OSI-approved licenses"

2. The Open Source Definition as clearly defined by OSI (and accepted as the only consensus definition in the world) does not include OSI-approved as a factor. So, a license can be Open Source without the OSI having done the active approval to validate that they see it as Open Source.

OSI-aligned might be a better term.

Which is why I am using "open source". You are free to use "non-OSI source" for your stuff.

The only thing you are saying is that "open source" sounds nicer and you want it for your term instead. Too late.

"Open source" was used to refer to source-available material for decades before it ever OSI ever existed. This is the reason why OSI can't get a trademark on the term -- UTPTO won't let you trademark a generally descriptive term already in use.

That's nice, but just says that it used to mean something else and you are bitter about it.

AWS, Google, and MSFT are among the top sponsors of OSI*. They are not neutral arbiters in the OSS space.

[0] https://opensource.org/sponsors

The lack of ethics is what poisons the relationship between big tech companies and open source software and OSI was founded exactly for that, to strip the ethical principles of free/libre software and focus solely on technical and economical output. Thanks for remembering us who are OSI sponsors.

Do we really need a ‘legally clear’ definition of open source and free software? Both look like a common term. Want something you could claim ownership over and exact your specific meaning — pick some proper name like ‘Apache license’ or ‘lgpl license’ — pretty unambiguous.

The Apache are a group of culturally related Native American tribes in the Southwestern United States. I'm pretty sure none of their tribal authorities have given you a license for your software.

Are you unable to use any of the several free search engines available on the web, or are you just making a point?

Either way, this is relevant: https://www.apache.org/apache-name/

Talking about categories of things are useful. This is like saying "Do we really need a term for two-wheeled vehicles driven by pedals? Just say you have a Cannondale or a Schwinn."

>Do we really need a ‘legally clear’ definition of open source and free software?

The FSF has gone to great lengths to give a legally clear definition of free software. Something like the SSPL is not free software because it restricts Amazon's freedom to use the software. Stallman has always had an ideological goal when it came to Free software.

Not just Stallman, but OSI was founded with an ideological goal too. It's goal is to strip ethics from software development and focus solely on technical and economical output for companies while Stallman and FSF had focused exactly on the ethics of software development.

If we don't want the idea to be gradually chipped away until it's meaningless then yes, we need to draw a line in the sand and say that open source can have these restrictions but not those restrictions. And thankfully we already have that: the OSI's definition, Debian's DFSG, and the FSF's Free Software Decision are all substantially equivalent and represent a clear consensus on what is and isn't open-source; one that a lot of care and thought have already gone into.

> Using the term "open source" without any definition is useless. If we can't agree on a definition, it's impossible to know if we're actually talking about the same thing.

Here’s the Cambridge Dictionary’s definition of open source: “Open-source software is free to use, and the original program can be changed by anyone.”

Merriam-Webster’s definition: “of software : having the source code freely available for possible modification and redistribution”

>If we can't agree on a definition, it's impossible to know if we're actually talking about the same thing.

We can certainly be more specific as to OSI Open Source rather than just Open Source? Wouldn't that solve the problem? We dont have to define every single term to have no ambiguity, but we can certainly have specific term for specific purposes.

You are correct that two people discussing "open source" should agree what they mean. You are not correct that what they agree on needs to match the OSI's definition. This issue can be solved by just asking someone what they mean.

Just to be a bit snarky. The GP is correct. The problem is they are technically correct. And we all know what is said about this type of correct.

> Given that OSI is the only body I know of with a clear definition

If this is the only definition you're aware of, then apologies, but you're not the right person to be attempting to drive this discussion.

If you're aware of the many others but do not consider them 'clear', then there is something else going on here, and I am starting to wander about agenda.

Ok, could you give the other (clear) definitions you are aware of? That might be actually useful here. Thanks.

"Open source" -- meaning the source is available and you can make modifications. You know, precisely what was in use before the OSI came along.

Sure, that's a starting point. What is missing is who "you" is and what can be done with the modifications.

If I say "source is available and you can make modifications, but only if you have red hair and live in Brazil", is this open source?

I think the common-sense definition that a lot people use, especially people not deeply invested in the movement, is something along the lines of "the source is available for everyone to examine and make modifications to".

It's not as specific as the OSI definition of course, and it doesn't say anything about redistribution and such, but that's okay and it doesn't really need to be IMO. Some vagueness is okay, and even with the OSI definition there is a huge difference between something like MIT and AGPL, both of whom are "open source".

A large majority of software developers and related professions use OSI's definition in my experience. Even the ones apathetic or antipathetic to the movement. The elements of OSI's definition have practical consequences for them.

People talk about permissive and copyleft licenses when the distinction matters. Covering both doesn't make open source vague.

Alternative definitions restrict the freedom to run the software at all frequently. Never mind modifying it.

Ok, thanks, that's one. But you said there are "many others"? Many other clear definitions of "open source", you seemed to be saying.

Yes, and butterfly is a fly made of butter. English words mean what people understand them to mean.

The only thing you are saying is that other people tried to use "open source" for something else and completely and utterly failed to convince people. Thus, today "open source" means what it means.

“Open source” has long been used used to describe research of publicly available information. A quick search in google books shows references back to at least the 70s. This meaning is consistent with the “source available” usage today.

Reply on news.ycombinator.com