Hacker News Reader

Unless you open-source your code or describe it publicly, I can't tell what algorithm you use.

So there are 2 possibilities here:

1. Your algorithm really requires years and millions of dollars of research to develop, in which case you have little to worry about - your competition is years behind you and probably doesn't have millions of dollars to spend on re-developing the algorithm. Trade secrets will work for you just fine.

2. Someone can quickly and cheaply come up with the same algorithm independently from you, in which case you don't deserve patent protection in the first place.

When it comes to patents, everyone brings the "super hard, expensive to develop" case, which carries the assumption that they're so smart that no other persone on the planet can come up with the same idea.

The reality is that even the patents that survive in courts and reap millions of dollars in damages, are laughably trivial and are a result of routine work of software developers, not to mention the vast number of patents that were granted and then stricken down during litigation, but only after both sides spent hundreds of thousands of dollars on said litigation.

> Unless you open-source your code or describe it publicly, I can't tell what algorithm you use.

But how do I sell it to someone without opening the code or describing it to them? With trade secrets, I'm forced to go into the business of building military radios (or whatever). What sense does it make for a small shop of wireless technology experts to go into the manufacturing business, learning how to manage supply chains, support end-users, etc?

That's the very problem patents were designed to solve 200+ years ago when they were introduced. To allow a separation of design from manufacturing. To allow the guy who invents the telephone to sell it to someone without going into the business of making telephones.

Think of other property abstractions, like stock. Stock allows separation of the roles of investor and manager. The CEO doesn't have to come up with the capital to buy the company, and the shareholders don't have to figure out how to run it. That's a useful separation of concerns. Patents allow separation of the roles of designer and manufacturer. ARM can focus on designing CPU's and patenting the novel features. They don't have to get into the business of manufacturing all of the different products that might contain their CPU core.

I'm ok with algorithm patents. It seems like the vast majority of abuses involve business method patents. I'd like to see those abolished.

Software patents exist only in the US, those business will do exactly what they've always done in the 90% of the world that doesn't have them.

Software algorithms are patentable, with more or fewer restorations, in the U.S., China, The UK, the EU, Japan, South Korea, Australia, and other countries.

Nonsense, the scope of software patentability is so limited outside the US that they do not influence business at all. http://en.wikipedia.org/wiki/Software_patents

Doesn't copyright law already cover this?

Copyright protects software, patents protect designs.

For some types of software, say operating systems, the hard work is writing all the lines of code. No device driver is ground breaking, but all those devices have to have device drivers and someone has to write them. The product you end up selling is those lines of code. Copyright protects those specific lines of code you wrote.

For other types of software, the hard part is figuring out how the software should work. That's where all the research dollars and engineer time gets spent. The actual lines of code are often an afterthought. If someone saw the code, they could easily figure out how it works and write an independent implementation. They can do that because writing the lines of code is the easy part. Copyright only protects the lines of code, so it can't help here. That's where patents come in.

In both cases, the law facilitates a division of labor. Copyright lets one company write an operating system, and sell it to others. Copyright creates and protects the subject matter of that transaction. Patents let one company design an algorithm, and sell it to others. Patent law creates and protects the subject matter of that transaction.

Now, I don't think the law should be concerned with protecting particular business models. However, it should be concerned with facilitating the division of labor. Indeed, that's one of the key purposes of property law. And that's why I think software patents have to exist in some form. Because I think it's good to be able to separate the process of design from the process of implementation, and without being able to protect design you can't do that.

I should point out that ARM is a very good example of this design/implementation separation. What do you think ARM uses to protect its designs?

Your argument works only if those who patent stuff are (i) better stuffers than most, and (ii) wouldn't have stuffed without patents. This is difficult to measure, because patents hinder those who want to build on your stuff (you want a cut, so it's more expensive to them). For instance, if ARM didn't have the monopoly over its designs, maybe people would step up and do the same work for free. The best analogy I can think of is free software, most notably GNU and Linux.

On the other hand, maybe ARM wouldn't have designed any chip at all. I don't think that would be a problem however. Technology tends to happen no matter what. Independent inventions of non-obvious stuff at roughly the same time are common. I don't know why, but it seems that when an idea is "ready", it just pops out of some earthling's head. (Citation badly needed, please.)

The idea to directly reward innovators is seductive, but I think it puts too much focus on the individual. Society as a whole doesn't need to reward something that would happen anyway. That would be a waste of resources. Not to mention the hoops you have to jump through to maintain an artificial reward system (the legal side of patents an copyright is quite expensive).

Your argument assumes that the world of "stuff" resembles GNU/Linux. Very broadly useful things that lots of people have the expertise to contribute to, where the costs of contribution are low.

Most stuff doesn't look like that. Nobody is designing transmission power control loops in their free time, nor are they designing chemical processes to remove impurities from natural gas before combustion. You gotta pay someone to do that, and it's very expensive to do so.

Also, patents are often construed as a reward for innovation, but I don't think that's the only way to look at them. I think a better way to look at them is like other abstract types of property, e.g. stock in a company. Patents, like stock, allow division of labor and specialization. Nobody is designing high-performance CPU cores in their spare time, but Samsung could pay someone to do it. Without patents, they'd keep the design a trade secret, and NVIDIA, Apple, etc, would pay someone to make their own designs and keep those as trade secrets too. Patents (and copyright), allow a single company, ARM, to specialize in designing CPU cores, by creating property which can be the subject of transactions between ARM, Apple, Samsung, etc. In such a scenario, the patent isn't a "reward" for innovation, but simply a legal abstraction that makes certain types of business arrangements practical.

> Your argument assumes that the world of "stuff" resembles GNU/Linux. Very broadly useful things that lots of people have the expertise to contribute to, where the costs of contribution are low.

I don't think I have to assume that. Individuals have to limit themselves to such contributions, but companies can make the bigger ones. The question is, would they? I think they would, because I doubt many such investments cannot be paid for without a (temporary) state granted monopoly.

(Now, if you abolished patents overnight, that could spur a sense of loss, which would indeed have a temporary chilling effect on innovation.)

> In such a scenario, the patent isn't a "reward" for innovation, but simply a legal abstraction that makes certain types of business arrangements practical.

Then, I would investigate the value of such business models. Specialization is good, but I think the additional specializations allowed by patents are hitting diminishing returns.

I would also investigate other means to enable such business models. If they can be achieved without patents (your example suggests trade secrets would work too, if Apple, Samsung, etc signed non-disclosures agreement to the same company), then that's one more argument against the patent system: between the patent office, the difficult suits and counter-suits, and the lawyers that work on this full time, the sheer organizational cost to society is substantial. If trade secrets can do the same more efficiently (I think it can in your example), then I say let's kill the unnecessary work!