[parfe]

Mueller was correct only because the jury had no other option on question 1A.

The judge instructed the jury that APIs were copyrightable. He did this to prevent the certain appeal of his decision on APIs from forcing another jury trial which would rehash the same testimony.

Mueller was "correct" because he didn't have to make a risky prediction on the jury trial.

In fact, Mueller has been wrong, repeatedly. He predicted Google would settle. He expected the case to be resolved months ago. He gives that as a reason for hiding his employment with Oracle, a blatant conflict of interest while he offered expert commentary on the case.

He predicted the jury would find against google's fair use defense for question 1B, then blames the judge for not issuing the correct instructions.

He's basically a huge anti-google marketing campaign.

Ask him how that Linux GPL infringement issue that was going to ruin the android marketplace has been going? Nowhere you say?

edit: Here's some context, but I regret having to link to his garbage blog to provide it:

Fake emergency ginned up by Mueller regarding distribution rights being revoked from Android vendors: http://www.fosspatents.com/2011/08/most-android-vendors-lost...

Fake and inaccurate licensing issue ginned by by Mueller regarding the "viral" GPL infecting[1] Angry Birds forcing Rovio to release the source: http://www.fosspatents.com/2011/03/googles-android-faces-ser...

[1] Hint, GPL licensing does not work that way.

[stock_toaster]

I am curious that if apis are solidly found to not be copyrightable, would this have any impact on gpl3's 'anti tivo' exclusions and/or gpl vs lgpl linking exceptions?

It would seem to me that providing an api compatible 'shim' layer would then not fall under copyright if apis are not copyrightable. Although linking is clearly, in my mind, the creation of a derived work.

My understanding of the many facets of copyright law is not very extensive though, so enlightenment is definitely welcome.

[wmf]

This has been discussed on and off over the years. Even if APIs are not copyrightable, it has been argued that using an API whose only implementation is GPL creates a derivative work. Clean-room shim header files sound like an admission of bad faith but they may be legally defensible.

[bad_user]

     would this have any impact on gpl3's 'anti tivo' 
     exclusions 

When taking a GPL v3 implementation and distribute it, there is no license that grants you that right, other then the GPL v3 itself.

You are basically mixing here 2 different things: the implementation and the interface. GPL v3 affects the former, while APIs not being copyright-able affects the later.

     It would seem to me that providing an api compatible
     'shim' layer ...

As long as your software is distributed along with the GPL v3 module, then the copyright of the GPL v3 module applies and the only way to distribute that module is to respect the GPL v3 clauses, because again, no other license or law gives you the right to distribute that GPL v3 module, other than the GPL v3 license itself.

What you can do is to allow the user to install the GPL v3 module by himself. This works as long as the GPL v3 module is replaceable. For instance, say that your software uses an SQL Database and that the preferred RDBMS recommended to users is a GPL v3 licensed database. You could even provide scripts that automatically install/configure that database. This wouldn't mean that your software has to be GPL v3 compatible, because the database itself is replaceable and it is also not distributed directly with your software.

Another example would be for instance the usage of GNU Readline ... your software could use whatever Readline implementation it finds on the user's system through dynamic linking, and as long as it works with other implementations, then there is no hard dependency and you've got no problems.

However in the case of Tivoization, then direct distribution of the GPL v3 package is involved. And you simply cannot do that, no matter what API layer you've got between that package and your software.

     Although linking is clearly, in my mind, the creation of a derived work.

Not necessarily. Linking, either static or dynamic, is an artifact of the way we build software, being basically just a protocol through which different software components can communicate.

The context is very important here. Let's say that MongoDB would be GPL v3 licensed (with no exceptions), there wouldn't be any other implementations and your software couldn't function without it. Because Mongo's APIs are unique (not a standard, no third-party implementations), would this mean that your software is derived from MongoDB? Yes it would, simply because without MongoDB it couldn't function.

On the other hand, if your software relied on MySQL, would your software have to be GPL v2 compliant? No it wouldn't, because you could just replace it ... however, in the case of MySQL the actual client library that gets linked to your software is also GPL v2 and has its own unique API that's relevant to MySQL only, so in such a case you have to find a way to dynamically link to that client, such that your software can continue to function even if the MySQL client library isn't installed.

But even so, this is about "usage", not "distribution". The distinction is important here, because if you want to distribute MySQL alongside your software as a single package, then your software must be GPL v2 compliant, no doubt about it.

Btw, I'm not a lawyer, this is just my understanding of how the GPL works.

[0xABADC0DA]

'' Because Mongo's APIs are unique (not a standard, no third-party implementations), would this mean that your software is derived from MongoDB? Yes it would, simply because without MongoDB it couldn't function. ''

According to the recent EU ruling your program is derived from MongoDB's API. But to prevent 'monopolization of ideas', MongoDB APIs are not under copyright so your software is not a derived work. In addition the ruling voids license clauses with conditions on how you run the software past normal use, so clauses like 'except you can't clone the API' or 'can only link with approved software' or 'have to publish the signing key' are unenforceable.

Since it's not a derived work and GPL linking restrictions are not enforceable, you simply accept the GPL and distribute the .so and source for that code along with your closed-source program and use it. You still have to publish your changes to the GPL code itself, like LGPL, but that's all.

This is the flip side of the API-copyrightability coin.

[yummysalad]

anti-google or not he correctly predicted that the jury would find that Android infringed on Oracle's copyright and that is my point. whether I agree with the jury or not or if I hate or Mueller is not the issue.

[yohui]

It's important to be right for the right reasons. All else equal, parfe's charges against Mueller suggest reason to discount Mueller's credibility despite occasional, coincidental victory.

[yummysalad]

in this case imo Mueller has more credibility than parfe (no offense). Why? because Mueller's opinions were backed by facts. parfe's opinions were supported by ad hominem attacks.

[rbanffy]

I think it's fair to ignore the expert opinion of both Oracle and Google employees on this matter.

It's, however, disturbing when Mueller pretends to be a neutral part, because, quite simply, he's not. The fact he's occasionally right doesn't make him honest. Or trustworthy.

[kprobst]

Just out of curiosity, how is that different than, say, Rob Weir's (IBM employee) opinion taken pretty much as fact by everyone on the ODF/OOXML conflict, or for that matter why do we accept Pamela Jones' analysis and opinions if she is so clearly biased against Microsoft?

[rbanffy]

Facts are facts and opinions are opinions. I don't see a reason to confuse them at all.

However, Mueller's opinion has been considered to diverge from reality in very specific ways, always favoring whoever is paying for it at the time and always being presented as an unbiased expert opinion.

[wtallis]

Having an opinion doesn't itself discount an opinion. Being biased or unreliable is reason to discount an opinion. It's important to remember that those are separate concepts.

Bias has nothing to do with how strongly you believe something to be true, or how hard you try to spread that opinion. Bias is nothing more than the extent to which your opinion differs from reality. If you consistently and vehemently insist that the earth is round and not flat, you aren't biased, you're right.

And regardless of bias, you have to also take into account the specific evidence and reasoning provided with an assertion. PJ in particular is pretty good about explaining her predictions, so you can't dismiss her solely due to her bias - you also have to find fault with her reasoning.

[zerohp]

You're misunderstanding what ad hominem is. parfe clearly attacked your assertion that Mueller has been correct so far, and supported that position with multiple objective facts.

You have not countered his argument with a similarly well-supported position, which is why you're being down-voted.

[yummysalad]

>He's basically a huge anti-google marketing campaign.

>his garbage blog

ad hominem attacks against Mueller. not me.

[ktizo]

Saying that someone is an anti-google marketing campaign, based on them being paid by google's opposition, is an accusation that may or may not be true, but is hardly a personal insult and is not a fallacial argument as it pertains directly to the matter at hand. The actual truth of a statement is not the requirement for it being considered a logical fallacy.

As to calling his blog garbage, his blog is the method by which he is communicating his views and so it is a judgement of quality of his writing, not of him directly.

[edit] also a personal insult, such as calling someone a complete and utter muppet, can be a perfectly reasonable point to make in argument as long as it is particularly relevant. Which judging from your commentary so far, is something you may wish to keep in mind.

[revelation]

So your point is someone made a correct prediction? On a 50/50 chance? Well, duh.

[losvedir]

To be precise: two options doesn't imply a 50/50 chance, necessarily.

For example, if everyone thought that Google would win on 1A, and someone predicted Oracle would, and then Oracle did, then that's doing better than 50/50. (I'm not sure if that's the case.)

See also: Monty Hall problem.

[yummysalad]

parfe's judgement is clouded by his hatred of Mueller. that's the point.