Hacker News Reader

Comment by rayiner | Hacker News Reader

It depends entirely on your field. I spent most of my career as an engineer working on R&D for software that would likely sit in a baseband in a production implementation. We spent millions of dollars building it, and nothing comparable is publicly available. The secret sauce in a baseband is definitely the sort of thing that's in short supply. And it's software.

That said, the patent office may not be qualified to distinguish between something that someone whipped together and something that took a team of PhDs years to invent. And maybe that problem is an intractable one.

View on HN

> We spent millions of dollars building it, and nothing comparable is publicly available. The secret sauce in say Qualcomm's baseband is definitely the sort of thing that's in short supply.

There's certainly short supply of the secret sauce in Qualcomm's baseband.

Your team spent millions building it, and, in order to give a sufficient incentive for teams to do similar things to that, the law should probably give some protection so that Qualcomm can make a profit from it. But you're conflating a bunch of things here:

1. Qualcomm knows the product and has the engineers. Qualcomm is uniquely positioned to build usable reference implementations, chips, etc, and to charge for this service and for support. This requires no legal protection whatsoever.

2. Qualcomm holds copyright on the code base. This has nothing to do with patent law, and I think that very few people would dispute that copyright makes sense in this context. (Whether long-term copyright makes sense is a different story altogether. I tend to think that a decade or two would be more than enough to make Qualcomm's investment worthwhile enough that they would do it.)

3. Qualcomm holds lots of patents that prevent other people from spending the same millions of dollars to develop a competing product without taking significant risk of being sued. I, and many others, think that the law should not offer Qualcomm this protection.

4. As you said, the sauce is secret. I think the question of whether Qualcomm should be permitted to keep the sauce secret and how much the law should help or hinder this secrecy is worthy of at lease some debate.

> As you said, the sauce is secret. I think the question of whether Qualcomm should be permitted to keep the sauce secret and how much the law should help or hinder this secrecy is worthy of at lease some debate.

This is going in circles. You cannot patent something without disclosing it. So, either Qualcomm has a patent, and it isn't secret, or they don't have a patent (nor can they ever, after 18 months) and the argument is moot.

Yeah, in general stuff pertaining to software can be classified into one of four kinds of IP: copyrights, trademarks, patents, or trade secrets.

The date that matters for trade secrets with respect to patentability is the date that they are first published or offered for sale. I don't believe offering a binary for sale counts as offering the trade secret for sale, but insert usual disclaimer.

> This is going in circles. You cannot patent something without disclosing it. So, either Qualcomm has a patent, and it isn't secret, or they don't have a patent (nor can they ever, after 18 months) and the argument is moot.

Yes, but then software patents should require you to release a full reference implementation with full documentation under a free (BSD, MIT) license as soon as the patent runs out.

If we were going to enforce the same transparency that non-software patents give then they'd release the full reference implementation right in the patent.

Exactly — but for all practical measures, it does not differ if they release it in the patent, or only when the patent runs out.

Currently, they don't release it at all.

Additional regulation I'd like to see: Any neural networks trained based on data of users have to be completely public domain. If Google wishes to train recaptcha, they should either pay for people to train it, or give back to the public the value that the public created.

It does matter, because if the publish it in the patent I can start planning to build something with years before the patent expires, or time improvements I make on it for that expiration.

This to me is the sticking point. The biggest benefit to society for patents is: you get the formula (but can only use it once the patent expires). I would imagine that no one short of lawyers has ever bothered to read expired software patents. They are literal drivel and until they are useful to have disclosed, they should probably not be patentable, in my opinion.

patents should require you to release a full reference implementation with full documentation under a free (BSD, MIT) license as soon as the patent runs out

Why not as soon as the patent is granted?

Because that bill would be even less likely to pass?

It is indeed. You are arguing that the secret sauce is in the baseband, and yes you can't patent it if the competitive advantage is that it's hard to make so long as it is secret. It's novel, you don't need to patent it to make money as copyright takes care of that... And so to complete the circle patents aren't necessary, and don't benefit society in the scenario you present!

edit: you might bring up clean room engineering that AMD did in the 80s. Intel weren't effected, and are bigger than ever. And the pace of innovation is massive.

> 1. Qualcomm knows the product and has the engineers. Qualcomm is uniquely positioned to build usable reference implementations, chips, etc, and to charge for this service and for support. This requires no legal protection whatsoever.

Why should you or anyone else get to decide on how Qualcomm is monetising the product of their investment?

Qualcomm spent millions on developing this, nobody else should be able to decide this.

> 4. As you said, the sauce is secret. I think the question of whether Qualcomm should be permitted to keep the sauce secret and how much the law should help or hinder this secrecy is worthy of at lease some debate.

It is Qualcomm's property, they can keep it secret or open source it however they decide. I deeply reject the idea that government should have the right to order us to share our private property with the public.

> It is Qualcomm's property, they can keep it secret or open source it however they decide. I deeply reject the idea that government should have the right to order us to share our private property with the public.

Mostly I agree with you, but I think it is worth noting that this discussion is in the context of patent protection for software.

One of the requirements of applying for a patent is providing a reference implementation - I believe this is what the parent was referring to when they mentioned whether or not Qualcomm should be "permitted to keep the sauce secret".

Well yes, if that's what he meant then it's something different. There's no point in allowing patents when nobody can know what implementation is patented in the first place.

Why should you or anyone else get to decide on how Qualcomm is monetising the product of their investment?

Because Qualcomm is riding the coat-tails of a government-issued investment protection? You can't argue for anti-free-trade policies by using free trade arguments.

While perhaps true for Qualcomm, it is hardly the case for all agents producing valuable knowledge.

For the sake of argument, if Qualcomm had no such connection to the government and has created it's products entirely from private funds, why should the government or other third parties have a right to dictate their product?

>> 1. Qualcomm knows the product and has the engineers. Qualcomm is uniquely positioned to build usable reference implementations, chips, etc, and to charge for this service and for support. This requires no legal protection whatsoever. > > Why should you or anyone else get to decide on how Qualcomm is monetising the product of their investment? > > Qualcomm spent millions on developing this, nobody else should be able to decide this.

I said no such thing. I said that Qualcomm could do this, not that they must or that they should be required to.

>> 4. As you said, the sauce is secret. I think the question of whether Qualcomm should be permitted to keep the sauce secret and how much the law should help or hinder this secrecy is worthy of at lease some debate. > > It is Qualcomm's property, they can keep it secret or open source it however they decide. I deeply reject the idea that government should have the right to order us to share our private property with the public.

I don't think the government should, in general, require Qualcomm do share this secret sauce. But it just might be worth considering whether the government should require that the secret sauce that runs enormous numbers of devices (read: extremely high-value target if someone can break or pwn many of them) on wireless spectrum (which is a shared resource!) should be opened to independent auditors and security researchers as a precondition of being granted a license to operate on public spectrum.

Like I said, I think it's worthy of at least some debate.

Bet my team of PhDs would come to the same solution.

And there you have why your situation is no different.

Software is a dish, we are chefs, and techniques are our recipes and principles. You should no sooner be able to patent a chicken casserole than a baseband. We merely discover what is already true and turn it into code.

> Bet my team of PhDs would come to the same solution.

Maybe they would, but they didn't.

The point of the patent system is to incentivize people to publish their inventions. The secondary purpose is to encourage invention. Giving a temporary monopoly is not the greatest solution, but try coming up with a better one.

> Bet my team of PhDs would come to the same solution.

What is missed here is they might of come up with a better solution, but due to the risk of a lawsuit they never even try.

I don't follow this - why would anyone get sued for trying to come up with a different solution?

Because they may contravene, a vague overly broad software patent, that effectually locks up ideas, not implementations.

Because the current way patents in the US are viewed, is as ammunition to attack competitors (and potential competitors). :(

Open Source software largely handles what patents are intended to do. If someone makes some cool software, then someone else works out how that works. Then they develop code, which they publish under an open license. Then others contribute to it, or if that doesn't work then they fork the code and thus innovation AND publication occurs.

And as software gets more complex, you need to support it and build new features. That happens all the time without the aid of patents. In fact, when a patent is made, then sure its patented but so what? It can't be used for decades, and so only one party benefits and the idea that is beneficial to society stagnates.

This is not a zero sum game here. If you patent a medicine for 10 years, then people like Martin Shkreli come along and jack up the price of their product, nobody is allowed to offer cheaper medicine and literally millions die.

Or you have a situation where a non-lethal arrow is patented, someone independently comes up with the same idea, documents it and sells it, does to want to patent it and then gets sued to an inch of their life. Then you have an independently document invention, and no innovation.

The argument doesn't stack up.

Open Source accomplish exactly the opposite of what patents are for: instead of encouraging innovation by providing financial incentive (from temporary monopoly on selling said innovation) it discourages it because it basically ensures that you'll never make any money on your invention.

Maybe I'm missing something but there aren't many examples of truly innovative open source software - most of OSS projects even describe themselves as "Open Source alternative to [insert innovative commercial product here]"

Email isn't patented, not it the web, nor is TCP/IP, nor is any one of dozens of non-patented inventions you use on a daily basis. If TCP/IP was patented, do you think you'd see the Internet as we know it today? Would NetFlix, or Wikipedia or even Google would exist in the form it is now? I don't think so.

edit: removed the "that's ridiculous" comment. I'm sorry, you aren't ridiculous, nor is your comment. That showed you a great deal of disrespect, and I was out of line. I hope you can accept this apology. My disagreement with your argument should never have led to disrespectful comments.

(I just hit the up vote arrow)

I think some of my other comments have this language in them. That's unfortunate, it's just this is just the first one I noticed.

Actually, open-source is the exact opposite of what patents are intended to do - patents discourage trade secrets that die with their inventors by giving their holders a right to timed exclusivity (regardless of whether other people know how to copy them or not), in exchange for publicly documenting said trade secrets.

Open-source is essentially just a tool for collaborating on software more easily (ignoring the Free Software side for a sec), but which requires giving up propriety over the software.

Can you give any example of a trade secret that cannot be reinvented had it not been for patents?

Patents are hundreds of years old. In the fifteenth century markets and science was very different to what it is today. There is an argument that patents were helpful before the world evolved to freer markets. I mean, patents were invented around the time of feudalism, and nobody now considers THAT to be a good idea. An example used for the case of patents is James Watt's steam engine, an amazing invention. But in my view it is an example of an invention where innovation was deliberately stymied; sure he improved upon it greatly, but then again he prevented any number of other capable people from doing so and caused a monopoly. And ironically he could not have patented his invention had he not taken the original non-patented ideas of Thomas Savery and Thomas Newcomen, who originally worked out the general principle.

In fact, you only have to look at the most important invention IMO the world ever produced to see that patents would have stopped innovation entirely - the printing press, which predates patents by a century. Without the printing press, almost no modern innovation would have been possible. There were vast numbers of printing presses produced that allowed much more rapid dissemination of ideas and ushered in the current age. Had a patent been allowed on this innovation, then the modern day would look much, much more different. Even religion would have been different.

No, patents may discourage trade secrets, but you assume that they are the only mechanism for doing so, and you dismiss the fact that they aren't necessary to prevent someone else from discovering that trade secret.

You have a few conflated arguments here.

Open source is relatively orthogonal to the patent system. There is some overlap in that they both require disclosure, but the fundamental incentives are different, so you should expect different software to be developed.

Improvements to patents can be patented in their own right, by different owners. I agree that it is significantly more onerous than open source.

Scummy as Martin Shkreli was, he jacked the price up on a generic... Not something under patent. It just takes time/money to certify that new generic lines are manufactured to FDA standards, do he had a temporary monopoly.

Lawsuits against "infringing" software are relatively easy, and expensive to defend against. That is independent of whether the defendant patented the software or not. This is more the nature of our legal system, and one could argue that the patent system should handle this stuff internally. However, I think the legal system would be hard to improve upon without causing other problems, like making it difficult to challenge for legitimate infringement.

All in all, I agree a lot of questionable patents get issued. I just think the reality is far more subtle.

India decided to ignore patents on certain medicines. Pharmaceutical companies were initially outraged, then when they realised they had hit an immovable object they freaked out and started offering their drugs at very low cost. They are still making innovative products, but they distorted their market power and millions of poor Indian citizen's lives were immeasurably worse off for it.

Your final point about patent infringement... I dunno what to say. You've just explained why patents are inherently problematic, the patent system can't handle it internally because then they would be handling financial lawsuits in a non-transparent matter, and the rule of law would be in essence damaged. By making the patent office the approvers AND handle objections - that's not going to ever work. They'll almost always agree with their own decision. After all, they made the wrong decision in the first place!

Patents occur within the legal framework. If the patent system is hard to improve, then those other issues you mention are, to my kind, not immeasurably greater than the utterly broken patent system we gave today.

I don't feel the reality is as subtle as you talk about. I think it's actually as clear as day that patents distort markets, and I don't feel you've demonstrated that patents aren't orthogonal to open source software, except where it comes to financial incentives.

Open Source financial incentives are to build the best software that the market wants, and sell the code with backing support contracts and early access to new features that are shown in the open, with guarantees that the best people are working on the software; yet still the same time allowing maximum competition and innovation in the market whilst each company or organization strives to produce the most attractive offering.

Patents attempt to produce a monopoly on an idea and prevent anyone else from capitalising on it for decades. It assumes that nobody will ever want to form an invention unless they have exclusive control over that idea. That's self-contradictory - protecting a patent means that someone else WILL have that idea eventually and want to build a product around it, or even incorporate it into their product. They won't be able to, except at great cost. This clearly and IS stifling innovation. The rise of patent trolls and defensive patents are just two examples of how utterly and fundamentally flawed the entire concept of patents are.

This is a silly forum in which to have a long debate :). I will admit, I mostly agree that the vast majority of software patents should probably be invalid, and that the existing system leaves a lot to be desired.

A couple of important points:

> They are still making innovative products, but they distorted their market power and millions of poor Indian citizen's lives were immeasurably worse off for it.

No, most innovation in new drugs happen in countries with strong patent protection. There is a lot of development outsourced to India, but almost nothing innovative on the development side. That isn't to say India's policy doesn't make sense for them, but they are essentially just stealing from the US and other countries. Imagine if that were true worldwide 20 years ago. Health insurance would be a lot cheaper, yes, but a large percentage of drugs developed in the last 20 years would not exist. It's a trade off that maximizes a longer term return.

> infringement

Yeah, I was just pointing out that a lot of the patent system is designed to protect people who are being infringed by larger entities. Just because it is abused by others doesn't mean it doesn't serve a purpose.

> patents distort markets

Of course they do, but it is not necessarily net negative.

> demonstrated that patents aren't orthogonal

Yes, that is a very long discussion. However, it should be fairly obvious that the software developed by industry and the software available in the open source world is not the same. I would go so far as to say the open source world is an order of magnitude or two smaller than software from for-profit entities. There are different incentives for releasing open source software than starting a business, and patents are designed to make the latter more attractive.

> Patents attempt to produce a monopoly on an idea and prevent anyone else from capitalising on it for decades.

No, patents provide a temporary, licensable monopoly on an implementation. You don't patent the idea of holding paper together, you patent the concrete implementation and manufacture of a paper clip.

More importantly, they require disclosure of that implementation to the rest of the world. I think this gets lost when people talk about removing the patent system. The primary purpose wasn't to encourage invention, but to make sure people disclose it.

Agreed. For the record, I hope you understand that whilst I disagree with you, I respect your viewpoint :-) if I've been overly harsh, I apologise. I agree this is a long conversation and probably not entirely the right forum. Also, I just had an operation and in pain, and just took some heavy medication, so I'm going to have to bow out... My writing might suddenly appear more erratic than normal!

I think the biggest problem is people independently inventing things. Late last year, I thought for sure the box shadow drawing technique I spent several days developing from first principles was original. Turns out that, of course, it wasn't: the technique had been described a year or two before that and was written up on a blog.

Carmack's Reverse is probably the most well-known example of this, also in the graphics field.

It may still be optimal in some macroeconomic sense to encourage two people who independently invent an idea to race to patent it and to have the winner of the race sue the loser. But I think most people would consider this scenario unfair.

This is a problem with all patents though, independent invention is not a valid defense…

I think an even more stupid example was the xor cursor technique that was locked up by CadTrac, (later Quantel) and forced a rewrite of parts of X, and the Amiga.

https://plus.google.com/+RayCromwell/posts/dbipY1GJoGv

>The secret sauce in a baseband is definitely the sort of thing that's in short supply. And it's software.

I think some of the general attitude towards software patents has to do with CS's very close historical relationship with the math department, with many of the old CS advancements and algorithms having been done by mathematicians or at least math majors (if you count people like Knuth and Hoare) and in math there's always been a very strong culture of work-sharing as well as the general attitude that many discoveries were fated to have been made by someone (such as the contest to prove the theory of quadratic reciprocity), and there are many famous examples in history of different people independently discovering the same proof. It's this latter belief that causes people to believe that patents have a "chilling effect": when there are no more algorithms to be discovered, the only way to develop an effective product is to pay one of the first movers, also known as "finders-keepers", also known as rent-seeking.

Patents are awesome when they discourage stealing work, but bad when they discourage independent reinvention. And it's really the possibility for independent reinvention, not "triviality", that makes software patents a problem. Unfortunately just because a problem isn't simple doesn't mean that two different people won't find similar solutions independently. When someone gets a patent on the only effective way to do something, it can create a legal barrier that prevents some people from being able to do it at all, like using a Mac in the '90s with a decent mouse, or listening to music on an open-source device, or HTML5 <video>... who knows how much stuff hasn't been made because of MPEG-LA?

It actually depends on the idea. Einstein's theory of General relativity and special relativity are "ideas." So are my "ideas" for building the next facebook killer. One is more complex and "correct" than the other, and each "idea" is merely a description of things—a collection of infinitely smaller ideas. Descriptions can be detailed, wrong, or absurdly vague.

I have an idea for obtaining two apples. It involves going to the apple tree and obtaining another apple and adding it to the one I already have. Simple, but valuable idea because its constituent ideas are probably low cost (the apple tree has apples... find an apple tree... reaching for the apple is also a simple concept). Of course the value of this idea is the sum of its constituent ideas. So it may turn out that there are no Apple trees in the vicinity and no one knows where to find one. Then the value of the idea is inversely correlated with the cost of the constituent ideas (finding out who has an idea where there is an apple tree). The idea for physically obtaining the apple once you are at the tree is to physically reach your hand up to an apple and apply a sufficient force to remove it from the tree. It's a brilliant idea, but it is also cheap since 90% of people over 3 years old probably know how to do that.

Well, you say that the software wouldn't be written, but then I don't believe this. Just because someone wants a monopoly on an idea, doesn't mean it can't be useful to make someone else money. If the idea is valuable enough, then someone will make a product from it. Then someone else might improve it, or make it cheaper. That's competition, in a free market.

The funny thing is, if software patents were only for 5 years, then it's unlikely this would be a story. The market could be capitalised on still by first-movers, and then competition and those who want to contribute further to society could cool their heels, even study the designs, and move product for consumers, and provide additional jobs. Mass market distorting giants couldn't get unfair advantages and innovation would thrive. The technology would improve because then that same company could file improvement patents, but not stop the old patent from being used. Patent trolls would not be a thing, or at least a very limited thing and at the very least all easy processes would be documented.

Teams of PhDs would still invent things. We'd still have cool tech. Open source people would make amazing things. Matheticians would continue collaborating. The world of commerce would not come to an end.

Judging novelty or obviousness of software is an intractable problem placed on the patent office. There are too many people "inventing" diverse software too quickly. If millions of people carried gene labs around in their backpacks it would be an intractable problem for biology patents too.

How do patent officers judge novelty or obviousness? I mean, I find a loop obvious, but a non-programmer applying it to another domain might consider it novel, even if it's not to me.

The answer: they can't, it's all subjective to the examiner, and the examiner has an incentive yo eventually pass the patent. Then it has to go through an expensive court case where only a patent troll wins because they threaten lawsuits which some companies can't afford, even if the patent is bullshit, or nobody wins as the patent troll doesn't succeed, but no or inadequate costs are awarded to the defendant.

> I spent most of my career as an engineer working on R&D for software that would likely sit in a baseband in a production implementation. We spent millions of dollars building it, and nothing comparable is publicly available. The secret sauce in a baseband is definitely the sort of thing that's in short supply.

A lot of those millions of dollars are because you have crappy processes, crappy tools ("rational" ones for example), base-level engineers (with some exception) and stupid management

I bet most of your "secret sauce" is mostly application of Computer Science 101/201

I know, those softwares are complicated. But I'm not feeling sorry for most big companies who like to justify cost with their own crap dragging them down

(And yeah, I know, telecom protocols are awful)

That's why Android and Apple wiped traditional phone companies

You clearly have no idea what you are talking about. I am doing baseband SW in tier1 vendor, and we have no processes, crappy tools (maybe you consider git and clang a crappy tool?) and don't give a flying f##k about management.

...and this is funny that you consider advanced signal processing, numerical analysis, SoC architecture, discrete optimization a CS 101 - what a joke.

> maybe you consider git and clang a crappy tool?

No I don't, and good for you that doesn't work with bozos that like ClearCrap, oops, ClearCase. Apparently they evolved

> that you consider advanced signal processing

No, I don't consider that simple, but sometimes the frontend signal processing is split (DSP stays in one part and higher level processing in another part)

> numerical analysis, SoC architecture, discrete optimization

Well, SoC architecture is hardware, no?

Why the downvotes?

Google and Apple don't write their own baseband software.

Yes, and companies that have worked in that area have left the mobile phone marked (Qualcomm and Philips for example)

I don't understand what you mean.

Reply on news.ycombinator.com