LWN: Comments on "OSI board AMA at All Things Open"

Do they even see themselves how utterly ridiculous they are?

ssmith32 — Wed, 13 Nov 2024 19:27:54 +0000

A misleading and logically flawed argument.

Stating you should not call CAL open source, does not preclude differentiating between CAL, and other licenses. It just precludes calling it open source.

Do they even see themselves how utterly ridiculous they are?

excors — Thu, 07 Nov 2024 17:32:40 +0000

He accepts it won't work for some companies:

> We also have some non-goals:
> * Don’t worry about when or if today’s giant companies will join Post Open. The answer to "When will IBM come on board" may well be "never", because for such a large company, participation would mean a new USD$60 Million dollar per year fee (1% of USD$6 Billion). Instead, attract small and new companies with a free license, and grow them into paid license customers.
(https://postopen.org/how-post-open-works/)

...but I think those numbers are wrong - IBM's annual revenue is more like $60B, so the fee would be $600M per year. ($6B is IBM's recent annual net income, or their quarterly revenue from "software" alone (excluding "consulting", "infrastructure", etc))

And IBM isn't even a very big company by revenue (219th according to https://fortune.com/ranking/global500/ - the top companies are 10x higher), and it gets a higher proportion of its value from open source software than most companies, and it has reasonably decent net profit margins (~10%, compared to e.g Walmart's 2% which is typical for the grocery industry, meaning this fee would be half of Walmart's entire profits), so IBM is one of the better cases for this licence.

I think the non-goal is effectively excluding all companies in low-profit-margin industries, and most large-ish companies in high-profit-margin industries, and any small/medium company which hopes to either grow into a large-ish company or be acquired by one. So I find it hard to imagine _any_ company would ever agree to this. And without buy-in from companies totalling probably billions of dollars in revenue, there won't be enough money to fund the project.

Do they even see themselves how utterly ridiculous they are?

pizza — Thu, 07 Nov 2024 17:27:04 +0000

> Licence violation is widespread and enforcement is almost nonexistent.

That's only a problem for copyleft "Free Software". It's nearly impossible to violate (therefore there is little need to "enforce") the "permissive" licenses that the "Open Source" movement embraced.

(Congratulations; once again, Stallman's predictions have been shown to be accurate....)

Do they even see themselves how utterly ridiculous they are?

excors — Thu, 07 Nov 2024 16:16:01 +0000

Yes, but evidently his views have changed after decades of experience, since he now says Open Source has failed. I think his main arguments are: Open Source's most significant achievement has been to create wealth for proprietary software companies like Google and IBM. There are very few successful Open Source projects for regular users. Critical infrastructure is severely underfunded. Licence violation is widespread and enforcement is almost nonexistent. And there is no workable solution to those issues within the Open Source paradigm. So he's working on a new paradigm, which he has explicitly called "a radical idea". (I'm basing this on https://postopen.org/ and some interviews and talks he's given.)

I have the impression that the wider Open Source community wouldn't fundamentally disagree with those problems he highlights, but the little discussion I've been able to find about his proposed solution has a lot of criticism and almost no support. Most people seem happy to bumble along with Open Source as it is now despite its problems, with perhaps a few incremental changes, while he's trying to shake things up, so it's unsurprising when that causes friction.

Do they even see themselves how utterly ridiculous they are?

Wol — Thu, 07 Nov 2024 14:59:28 +0000

I'm not saying I disagree with his idealism. I just think - in this particular instance - it won't work. Businesses are as assorted as people, and while it would work for some, I don't think it would work for most.

Cheers,
Wol

Do they even see themselves how utterly ridiculous they are?

Karellen — Thu, 07 Nov 2024 14:40:07 +0000

I'm not sure I'd give Bruce Perens' idealism a score of 'nil'. Pretty sure he's put a few points on the board over the ~~years~~decades.

Do they even see themselves how utterly ridiculous they are?

Karellen — Thu, 07 Nov 2024 14:34:18 +0000

And I guess it's unsurprising [Bruce Perens] fell out with the OSI when his ideas are so radically different to Open Source.

This feels like an odd take. Are you familiar with the history of the Open Source Initiative and its Open Source Definition?

Do they even see themselves how utterly ridiculous they are?

jkingweb — Tue, 05 Nov 2024 14:04:57 +0000

Thanks. I wonder how much of people's displeasure stems from OSI conferring "approval" rather than simply affirming consistency with the Definition.

Do I like the terms of the Cryptographic Autonomy License? No. I don't even like its name. Fortunately I don't have to use it, nor use any software which employs it. But it does seem to be consistent with the Open Source Definition at least as much as the AGPL, so I don't really see any inconsistency on the part of OSI.

Open Source AI is open-washing by any way of looking at it

zack — Tue, 05 Nov 2024 13:51:14 +0000

I completely agree with you on nuance.

I've also participated, as a volunteer, in the OSAID process. And once it became clear that OSI was opposed to mandating training data access, I've "battled" (for lack of a better term) for either a two-term definition (e.g., "open weight" vs "open source") or an additional qualifier (e.g., level 1/2/3/4), depending on whether training data were open data/public data/obtainable data/private data. I regret having lost that battle too. (But I still think that OSAID is better than nothing, given the current state of the AI industry, and that it will play a positive role in upcoming regulations.)

Open Source AI is open-washing by any way of looking at it

paulj — Tue, 05 Nov 2024 13:38:20 +0000

I think that for any generally queryable LLM made available to the public, you must be willing to accept the user is able to access (at least) snippets of the original training data. That is, the sensitivity of access to the prompt will be equal to the sensitivity of access to the original data. You need to treat the two - prompt and training data - as equivalent in terms of access.

In practical terms, for such LLMs, I think it will be required to anonymise any sensitive data.

It could be there are other kinds of AI models that can not, of themselves, directly leak the input data. E.g., a model arranged and trained to classify, say, diseases. You chat to it with your symptoms, perhaps, and it spits out a disease, and only a disease. If the output layers can only a set of output symbols that is a distinct set from the training data and much more limited than the training data, you could argue it is "safe" to distribute that model.

However, still, the training data is encoded into the parameters. The model state is a compressed form of the input. For sensitive input data, you would /not/ want to bet that it is safe to distribute the parameters, just cause they end up selecting from a predetermined and limit output layer. You would /not/ want to bet that some clever AI-hacker will eventually figure out how to backward-engineer some (or more) of the training data from the parameters.

Do they even see themselves how utterly ridiculous they are?

Wol — Tue, 05 Nov 2024 12:58:28 +0000

> and a second licence for larger businesses which requires royalties of 1% of the business's entire revenue paid to the Post-Open organisation.

So a large business that has nothing to do with the software industry will end up paying far more than a "small" software house, in return for much less benefit ... and depending on the definition of "revenue" a licence may well be out of reach for companies in low-margin businesses ...

That doesn't sound a sensible business model at all. "Idealism, meet reality! Score, reality 1 idealism nil".

Cheers,
Wol

Open Source AI is open-washing by any way of looking at it

lmb — Tue, 05 Nov 2024 12:34:22 +0000

I don't want to equate whether opening the data up is good or bad directly either. I, too, have been involved in this topic for three decades, I realize the truth is rarely pure and never simple.

There can be good reasons for data not to be made public or released openly. Medical, safety, personal privacy - and yet the AI/ML systems trained on them serve vital functions for society. Those may only be visible to officially appointed & chartered inspectors, for example.

Even the FSF acknowledges this - nonfree systems can still be just.

No, open/free/public data sets are not the only way to respect user rights.

Where the OSAID falls short in my opinion is insisting that those systems are "Open Source AI" - not everything can be open, and that's fine. (Or they could, indeed, say they're Open(tm) _only_ if they fall under such exempt regulations and are indeed independently reviewable.)

The *default* for an "Open Source AI" system should, in my book, indeed be open data. (And it'd be helpful if OSAID specified terms one could comply with.)

Again, my gripe is OSI going with such an all-encompassing term and indeed claiming their definition covers all components comprehensively - from a very prominent position with a lot of power and influence.

I think their OSAID 1.0 should have been more nuanced & differentiated and mostly stick to the parts we do understand reasonably well. And deliver something we can actually implement in practice. This, to me, reeks of preempting regulatory decisions, and/or marketing reasons.

I don't want in-fighting while the absolute exploitationists rejoice, either. But the OSI started the overreach with a reductionist definition, and also claims the high ground of authority - they get judged accordingly. They don't get free cheerleading.

The fact that so many assumed-to-be-well-meaning people see this as a potential erosion and open-washing clearly shows they've not produced something that is clear enough, if that truly isn't their intent.

(I also know you can't ever produce anything that is 100% proof against misinterpretation by malicious actors, but the folks whom I've seen voice criticism don't tend to fall into that camp.)

Open Source AI is open-washing by any way of looking at it

zack — Tue, 05 Nov 2024 12:18:19 +0000

> The Software Freedom Conservancy (SFC) however does seem to have their act and vision together.

Disclosure: I co-authored SFC's aspirational statement on LLM-assisted programming. As such, I am very near to SFC's position in this general space.

But note that about data training in OSAID, what SFC actually says is "I [bkuhn] truly don't know for sure (yet) if the only way to respect user rights in an LLM-backed generative AI system is to only use training sets that are publicly available and licensed under Free Software licenses. [...] My instincts, after 25 years as a software rights philosopher, lead me to believe that it will take at least a decade for our best minds to find a reasonable answer on where the bright line is of acceptable behavior with regard to these AI systems." And he is spot on.

The point that I'd like to highlight here is that once you start looking at the details (legal, strategic, philosophical, etc.), the data issue in AI/ML is quite complicated. Trying to simplify it down to require-data=good, do-not-require-data=bad is not going to serve us well in the long term.

Don't get all the hate

NYKevin — Tue, 05 Nov 2024 03:27:31 +0000

I disagree. The term "hate," in this context, is shorthand for "hateful rhetoric." It is not a claim as to anyone's actual emotional state, it is merely a claim that the words used are excessively charged and negative.

Which they are. You do not need to impugn people's motivations to argue against their interpretation of "open source." That is a pure ad hominem attack which has no place in polite discourse. Dismissing it as "hate" is an entirely reasonable and proportionate response.

Do they even see themselves how utterly ridiculous they are?

NYKevin — Tue, 05 Nov 2024 00:12:52 +0000

After skimming the article and spending far more time than I should have clicking around this thread[1], the problem appears to be related to these two sections of the license:

> 3.1. Permissions Granted
>
> Conditioned on compliance with section 4, and subject to the limitations of section 3.2, Licensor grants You the world-wide, royalty-free, non-exclusive permission to:
>
> a) Take any action with the Work that would infringe the non-patent intellectual property laws of any jurisdiction to which You are subject; and
>
> b) Take any action with the Work that would infringe any patent claims that Licensor can license or becomes able to license, to the extent that those claims are embodied in the Work as distributed by Licensor.
>
> 3.2. Limitations on Permissions Granted
> The following limitations apply to the permissions granted in section 3.1:
>
> a) Licensor does not grant any patent license for claims that are only infringed due to modification of the Work as provided by Licensor, or the combination of the Work as provided by Licensor, directly or indirectly, with any other component, including other software or hardware.
>
> b) Licensor does not grant any license to the trademarks, service marks, or logos of Licensor, except to the extent necessary to comply with the attribution conditions in section 4.1 of this License.
>
> [...]
>
> 4.2. Maintain User Autonomy
> In addition to providing each Recipient the opportunity to have Access to the Source Code, You cannot use the permissions given under this License to interfere with a Recipient’s ability to fully use an independent copy of the Work generated from the Source Code You provide with the Recipient’s own User Data.
>
> “User Data” means any data that is an input to or an output from the Work, where the presence of the data is necessary for substantially identical use of the Work in an equivalent context chosen by the Recipient, and where the Recipient has an existing ownership interest, an existing right to possess, or where the data has been generated by, for, or has been assigned to the Recipient.
>
> 4.2.1. No Withholding User Data
> Throughout any period in which You exercise any of the permissions granted to You under this License, You must also provide to any Recipient to whom you provide services via the Work, a no-charge copy, provided in a commonly used electronic form, of the Recipient’s User Data in your possession, to the extent that such User Data is available to You for use in conjunction with the Work.

Note also that "Recipient" is defined in a way that is similar to the requirements of the AGPL (i.e. it includes people who interact with the software over a network).

In English:

* The CAL license is both a copyright license and a patent license. So we need to analyze it like a patent license, and not just like a copyright license.
* If you allow an end user to interact with CAL-licensed software in some way that generates data, you must allow the end user to obtain a copy of their data, in a format that can be directly used by the software, and you must not modify the software in such a way that the user can't actually run it on their own private copy of their data.

Bruce's argument, as far as I can follow it, appears to be that the people who made CAL intend to use software patents to enforce the data availability provision as applied to their particular use case. In other words, they are not merely requiring that software based on the CAL-licensed code comply with this data availability rule, but are instead attempting to impose this requirement on all software that interacts with their (decentralized?) system, regardless of where the code came from. Perens also argues that this could be much more straightforwardly accomplished by simply requiring participants in this system to sign a contract relating to user data.

I'm not thrilled with the use of software patents for this use case. But I'm also not entirely convinced that this is a problem specific to CAL. This looks a lot more like a "software patents are evil" problem than a "CAL is not OSD-compliant" problem, at least from where I sit. Other participants in the thread pointed out that most other FOSS licenses (which mention patents at all) have similar "we are only licensing the patents that would otherwise be infringed by verbatim distribution" clauses, so it is rather difficult to argue that CAL violates the OSD on that basis, without then concluding that many long-accepted licenses also violate the OSD.

[1]: https://lists.opensource.org/pipermail/license-review_lis...

Don't get all the hate

ballombe — Mon, 04 Nov 2024 21:44:18 +0000

It is still not appropriate to qualify any of this as "hate". It is not.

Don't get all the hate

lmb — Mon, 04 Nov 2024 19:02:23 +0000

I think the observation that there are systems that are proclaimed to be "open" that don't ever meet the OSAID bar is valid.

And yes, OSAID *is* better than nothing.

I disagree with the assessment that this means we should not voice criticism to the term, nor that doing so is harmful.

OSI chose to use a very comprehensive single term with zero differentiation and significantly lower standards. They *could* have done it differently. Same with publishing a definition as "1.0" that actively asks for industry endorsement. But tell me, how would you actually comply with it? Under what terms would you make all that extra info available?

They position themselves as an authority and *the* steward. Their results get evaluated according to those claims, and their actions questioned for potential ulterior motives.

Calling that "infighting" ain't great, when folks see serious possible consequences of what they're pushing out. (e.g., the impact on political regulations.)

Open Source AI is open-washing by any way of looking at it

lmb — Mon, 04 Nov 2024 18:53:01 +0000

> Choudhury said that the OSI was spurred on by pending regulation.

I think this is the real motivation behind the industry stakeholders pushing this, and I'm not the only one to notice.

If OSI, the "authority" and "steward" of the Open Source Definition declares something to be "Open Source AI", surely it is, dear regulator? We don't need to make our sources open, it says so. See? Those more lenient obligations apply to us!

It doesn't even actually specify those "OSI-approved terms" all of the assets/components that aren't actual source code should be made available under.

But it gives the industry another fancy and easy-to-comply-with marketable label to slap on their products.

Everyone wins, except the public.

We'd not call something Open Source if it came with a description of the sources and where to go and (maybe) buy access.

I don't hate it, I get why OSI does it (it serves their stakeholders), I'm just hugely disappointed.

The Software Freedom Conservancy however does seem to have their act and vision together.

Do they even see themselves how utterly ridiculous they are?

excors — Mon, 04 Nov 2024 17:14:45 +0000

An old LWN article quotes and summarises the relevant bits of the licence: https://lwn.net/Articles/797065/

> The intent of all this language is relatively straightforward: if you are a user of an application (perhaps hosted on the net somewhere), you have the right to extract your data from that application to use with your own modified version of the code. Control of data is not to be used to lock users into a de-facto proprietary system.

From that article plus the Register one and some mailing list posts, it sounds like Bruce Perens didn't mind that specific licence's requirements, but he did mind the general idea of Open Source licences imposing requirements on data, largely because it makes the licences much trickier to comply with ("It's a good goal but it means you now need to have a lawyer to understand the license and to respond to your users"), and Open Source ought to be made simpler instead. He thought data restrictions should be completely out of scope for Open Source, and the OSI should reject CAL on that basis.

The OSI didn't have an existing policy on that, so it sounds like his arguments about field-of-use restrictions were stretching to find a technical justification within the OSD to reject it. And then he got increasingly frustrated when people didn't agree with him about the proposed policy or about that interpretation of the OSD, until he ended up quitting the OSI and accusing the board of conspiracy.

Incidentally, since at least 2020 he's been talking about what is now Post-Open (https://postopen.org/), apparently with the intent of supplanting Open Source by having a large number of software projects under a single new zero-cost licence for individuals and small businesses, and a second licence for larger businesses which requires royalties of 1% of the business's entire revenue paid to the Post-Open organisation. That organisation will subtract operating costs then divide the rest amongst individual developers (or their employers) in proportion to how widely each project is used and the number of lines of code each developer has written in the projects' Git repositories.

The goal is to make it simple for companies to comply - they don't have to spend any effort working out exactly which projects they use and how to follow all their different licences or pay for multiple support contracts etc, they just make a single payment to one organisation and that covers all their software - while also making them fund maintenance of the projects they rely on. Which doesn't sound like a bad goal in general, but his specific approach seems, uh, questionable. And I guess it's unsurprising he fell out with the OSI when his ideas are so radically different to Open Source.

Do they even see themselves how utterly ridiculous they are?

kleptog — Mon, 04 Nov 2024 15:33:15 +0000

You think there is no value in distinguishing between the CAL, which

> gives You unlimited permission to use and modify the software to which it applies (the “Work”), either as-is or in modified form, for Your private purposes, while protecting the owners and contributors to the software from liability.

and say, a Microsoft Windows license which doesn't even give you the source?

Do they even see themselves how utterly ridiculous they are?

jkingweb — Mon, 04 Nov 2024 12:48:04 +0000

I skimmed the license text and couldn't even identify what makes it not-open source. I'm really not understanding why I should be worked up about it.

Do they even see themselves how utterly ridiculous they are?

intelfx — Mon, 04 Nov 2024 11:08:18 +0000

> Unless you take the position that any non open-source license is proprietary and there is no grey area.

Uhm, yes?

Do they even see themselves how utterly ridiculous they are?

kleptog — Mon, 04 Nov 2024 10:27:49 +0000

So I thought I'd google to figure out what you mean, and end up with an article like this one which is totally useless. It does link to the license, but is basically an entire article full of quotes from people I don't know who disagree with each other, without even bothering to quote or reference the actual parts of the license that are a problem, so how can I make up my mind? On the other hand the OSI has an article which explains it well enough. It seems to be targeted at a very specific use-case.

People don't even bother trying to convince other people of their position anymore. Just a lot of "I think X" as if that is somehow enough to change my mind.

I asked ChatGPT for a summary why it's controversial and I can see it's unusual for a software license, I don't see anything that could label it "proprietary". Unless you take the position that any non open-source license is proprietary and there is no grey area.

Do they even see themselves how utterly ridiculous they are?

mirabilos — Mon, 04 Nov 2024 06:09:37 +0000

Huh, I hadn’t see that. It’s well hidden, and the language is obscure enough to qualify as deliberately confusing to people who are not English-language-native lawyers, too.

When they approved the “Unlicense”[sic!], an extremely badly worded combination of a (possibly not legal) PD waiver and a (definitely botched and not even remotely near working) attempt at a fallback licence that fails to actually licence anything of relevance, I was fed up enough, tbh. But this shows they lost the whole reason they exist in the first place.

Do they even see themselves how utterly ridiculous they are?

Paf — Mon, 04 Nov 2024 04:54:10 +0000

Can you say more about the issue with the CAL?

This is an unhelpful response from OSI

Cyberax — Sun, 03 Nov 2024 20:20:39 +0000

The OSI definition of "Open" is better than the status quo ante. Now it's clarified that the open models should make the weights available with clearly defined restrictions and document the training data set.

This alone will make it easier to filter out models that are not at all open (e.g. "OpenAI" or Facebook's llama).

The next step is to have the "Open Training Set" definition.

This is an unhelpful response from OSI

NYKevin — Sun, 03 Nov 2024 20:01:37 +0000

>> > "Nobody disagrees about the principles [behind the OSAID], where we see disagreement is implementation"
>
> That is a friendly sounding statement which I find rather nasty: it groups all the critics together and suggests they have a fundamental deficiency that is enough to not take them very seriously.

Frankly, I am not able to read that statement in the way you seem to be reading it. I do not understand "disagreement [about] implementation" to imply "deficient disagreement," nor to imply that such disagreements should not be taken seriously.

What I do understand is the background context: A large group of commercial entities (commercial-ish, in the case of OpenAI) have gone around claiming that their AIs are "open" in one way or another, despite flagrant violations of OSD#6 and various other parts of the OSD (to the point that OpenAI's product is just straight proprietary software, with no attempt to justify the use of the word "open" in their name whatsoever). What the statement you quote appears to be saying is that everyone agrees that those products should not be described as "open-source AI" - and under OSAID 1.0, they are all excluded, which is an improvement over the situation where these companies were going unchallenged in their use of this term.

This is an unhelpful response from OSI

IanKelling — Sun, 03 Nov 2024 16:47:38 +0000

> "Nobody disagrees about the principles [behind the OSAID], where we see disagreement is implementation"

That is a friendly sounding statement which I find rather nasty: it groups all the critics together and suggests they have a fundamental deficiency that is enough to not take them very seriously. But on the other hand, any criticism without that deficiency, well that is a disagreement on principles and you should not expect those to be resolved. The disagreements are important and this not a good response.

> She wanted to know how the community could work together to move forward.

> Scott said that the simple answer was to keep talking

You had a q/a, and chose to impugn your critics in various ways and did zero addressing of the substance of their criticism. That kind of "moving forward" is the kind where you can call it moving forward without moving anywhere.

Don't get all the hate

randomguy3 — Sun, 03 Nov 2024 16:44:23 +0000

sure, but there's a difference between saying "i don't think this was the right decision" and saying "only someone who was trying to destroy everything the open source movement stands for could have made this decision", or "there is no way anyone who made this decision could be taken seriously again". we've seen a lot of the latter ("they have lost all credibility" has popped up multiple times, even here on lwn where comments tend to be more moderate, with nothing to back up this assertion that the wider community's perception of OSI has actually changed significantly).

Do they even see themselves how utterly ridiculous they are?

kleptog — Sun, 03 Nov 2024 14:36:44 +0000

While I can't really comment on any of the specifics you give, reading some of the posts I was amazed by the idea therein that whatever the OSI comes up with is somehow relevant for the implementation of the EU AI Act and exemptions within there. That's not how this works.

The relevant definition of open-source for AI for the purposes of the Act is described within the Act itself (recitals 102-104). No, they don't require the providing of the training data. But more importantly, an exceptions for open-source AI are not available for any product/service placed on the EU market (Article 2(12)). So essentially unavailable for any commercial party like OpenAI or Meta. The idea that commercial parties could hijack the OSI process to secure themselves exemptions to the EU AI Act is just so far off the mark it's silly.

The exemptions also don't cover providing a summary of the training data and showing you complied with copyright restrictions. Which are probably the ones commercial companies are most interested in.

Don't get all the hate

zack — Sun, 03 Nov 2024 13:49:51 +0000

Very much that. And this way of debating legitimate policy disagreements (in this case on what "open source AI" means) is making the free software movement weaker against our actual opponents. Meta & friends, who have been calling "open source" AI systems that will never pass the OSAID bar, are eating tons of popcorn while watching this infight, and while lobbying regulators that *their* notion of "open source AI" is the real one. Good luck to us all.

Don't get all the hate

ballombe — Sun, 03 Nov 2024 13:01:21 +0000

> I don't get all the hate.
Disagreeing is not hate. Claiming otherwise is just a way to stifle debate.

Don't get all the hate

NYKevin — Sun, 03 Nov 2024 07:22:37 +0000

It's the modern state of politics. There is no gray. Either you're a good guy ("someone who agrees with my political goals") or a bad guy ("someone who does not").

Don't get all the hate

Cyberax — Sun, 03 Nov 2024 05:07:15 +0000

I don't get all the hate.

The FSF is free to come up with its own definition and push it. Nobody stops them.

And the OSI made it clear that the current definition is not the only possibility. Just like we have various copyright licenses with various levels of restrictions (from the super-restrictive AGPLv3 to the WTFPL).

Do they even see themselves how utterly ridiculous they are?

Shamar — Sat, 02 Nov 2024 21:44:08 +0000

> the board was open…

Not much. Several people were silenced, harassed or censored during the "co-design" process.

Julia Ferraioli paid a mental tool for trying to defend open source freedom from OSI's open washing goals:
https://archive.is/paD1W

But several other people felt the same and didn't dare to speak about it in public.

As for me, I was silenced for several weeks, and my posts were deleted because they debunked the OSI narrative:
see for example my reply to Carlo Piana (OSI board member) https://archive.is/JPSRX#post_3 that was removed from the forum: https://discuss.opensource.org/t/fsf-announced-basics-of-...

And something they don't even mention was the role Meta employees had to exclude training data from the requirements: here's where they admit the trick adopted (discovered by a user that was later silenced too) https://discuss.opensource.org/t/we-heard-you-lets-focus-...

For sure, the OSI has made itself obsolete, but too few people are aware of the alternatives such as https://opensourcedefinition.org/

Do they even see themselves how utterly ridiculous they are?

josh — Sat, 02 Nov 2024 08:28:04 +0000

> OSI has successfully made itself obsolete.

They did that a while ago, when they approved the "CAL", a proprietary license with usage resrictions that nonetheless can now masquerade as Open Source because it has OSI approval.

Do they even see themselves how utterly ridiculous they are?

mirabilos — Sat, 02 Nov 2024 05:17:53 +0000

yes, yes… the board was open… but those who actually drove the process weren’t and aren’t,
actively excluding voices all over, from FOSS experts to former board members, it didn’t matter,
dissenting voices are to not be taken for full by OSI.

> incredibly valuable to the work we have done

I bet the work was even more valuable to those…

> who ""may not have always affiliated themselves with open source""

And! Yes! Of course! Blame it on us poor people who don’t just…

> replicate the success we've seen in open-source software to AI"" and …

…insist on…

> simply translat[ing] the OSD to AI

… because we don’t have the “mind shift” necessary. Sure. That will be it.

I’ve seen enough. I barely skimmed the rest; OSI has successfully made itself obsolete.

Missing the target

josh — Sat, 02 Nov 2024 00:57:49 +0000

Also, in the very early days of Open source, far fewer people had the tools to work on software as well. Computers weren't as widely available, compilers and toolchains weren't as widely available, documentation wasn't as widely available...

Things get better over time. But not if you give up and decide to weaken your values to match what you think today's limitations might be.

Missing the target

ballombe — Fri, 01 Nov 2024 22:15:07 +0000

> ""The reality is that if only a handful of companies and a handful of governments have the resources"" to rebuild models, it is not a practical goal for open-source AI.

This is what openAI et al. want you to believe, but this is not true. There exist small models that can be rebuild
on commodity hardware. There are the ones which are important for the true purpose of open source AI : to allow to
understand and demystify how models are built.

Obviously big players need the process to be kept mysterious so that they can make whatever claim about it to judges, politicians and the general public. But the OSI must not become complicit in this.