Comment by rayiner | Hacker News Reader

Comment by rayiner | Hacker News Reader

MIPS is a great example of the kind of tech company that wouldn't exist without some sort of intellectual property protection. They're specialists in chip design--they don't make any physical products. The whole value of their business is their designs, which are protected by their patent portfolio.

In MIPS's case, intellectual property protection made it possible to compete on designing CPU's rather than designing finished products for those CPU's to go into. It allowed them to focus on chip design rather than fab design or manufacturing. And it allowed them to accumulate something of value in the company itself that could be sold later.

Would CPU's get designed without patents? Of course. But I don't think the sort of beneficial division of labor you see with MIPS and ARM would be possible without some sort of intellectual property protection. When you can't easily transact in something, that forces you to try and move every operation involving that thing under one roof. And it's not hard to make the case that those kinds of monolithic companies are worse for competition than patent litigation.

You can get those beneficial effects with trade secrets and NDAs, if _those_ are legally protected. All without a f*ing monopoly grant.

It is not practical to put an NDA on a mass marketed product.

Trade secrets can't be transacted in and NDA's don't protect against third parties. When you have hundreds of licensees, you can't bet the farm on all of them never accidentally or intentionally disclosing. Nobody would ever buy such a fragile property right.

The "monopoly" aspect isn't really relevant here. It's not like you accidentally stumble onto a MIPS patent. If you're infringing it's because you're copying the design.

*The "monopoly" aspect isn't really relevant here. It's not If you're infringing it's because you're copying the design.

Well that audacious. I'm will be less daring and say that with 500 patents, you are most certainly wrong about some of them.

Last time I said this, you brought up a 60 year old RF parent, but I'll give you another shot and say that I hav never seen a software patent worth protecting.

The RF patent was an example of the kind of patent I worked with, not a software patent specifically.

My example for a software patent would be the h.264 patent. That's not something a practitioner of ordinary skill would find obvious.

That's the entire f-ing point of patents. It's so they AREN'T kept as trade secrets. They are granted a monopoly in exchange for publication of it. It's the same reason the only prior art is something published, not a trade secret no one outside the company could have known about.

To carry on ismarc's thought, patents were only ever meant as a time-limited monopoly, to encourage and reward innovation.

Patents and copyright have grown to become much more defense structures and "keep out" signs, which I think unfortunate, but the basic premise is sound.

First, patents are tradeable, NDAs are not. Second, trade secrets are only as secret as the people knowing them, which is not very secure. Third, if you have a patent, you can discuss the technology without NDAs, which helps both in getting the technology accepted as well as in early-stage business negotiations.

Also, if anyone thinks getting rid of patents would get rid of litigstion, it wouldn't. An informal, ad-hoc patent like system would arise on top of NDA's and other contracts. Thse companies are always transacting with each other. Instead of suing Samsung for patent infringement, Apple would be suing for Samsung's copying Apple technology in violation of contractual obligations arising from Samsung's status as a supplier for Apple. Instead of patent litigstion in the ITC, we'd have litigstion over humongous iron clad contracts with suppliers in foreign countries.

Yes - but such a system would be optional and small/open-source developers wouldn't be afraid to innovate.

Small developers aren't immune to such activity. To the extent they do business with large companies, they can always be dragged in as co-defendants. "Apple: Samsung stole our technology and gave it to LittleCo, who were are suing separately for tortious interference with a contract."

The point is that if there really are so many entities just trying to abuse the system, they don't need patents to do it.

Small developers are best protected by patents, which simplifies using the judicial system to enforce their rights. Otherwise they would have to use a more messy system of contracts, as rayiner describes.

Suppose we stipulate that what you write is true. Does that good to society make up for the crippling costs and side effects of the very idea of ownership of ideas, and the current implementation of "intellectual property"?

Just as an example, a really tight "intellectual property" regime would have to have laws banning unapproved operating systems (to prevent copying) and laws banning unapproved encryption (to prevent transferring "IP"). Maybe the encryption would have official backdoor keys to allow for inspection of encrypted data. Possession of encrypted data would possibly become criminal, too. Then there's the money price of enforcement, which won't be trivial.

Is all that worth it so we can have a MIPS?

Surely some form of copyright protection would apply to the actual design of the chip to prevent people from ripping it off wholesale.

Would someone clean-room reimplementing a MIPS compatible chip be a bad thing, though? This isn't possible with patents.

Copyright protects only the most literal copying of a specific embodiment of a design. But when dealing with an embedded CPU, writing the VHDL isn't the hard part. The hard part is the circuit design. Compare this to a piece of enterprise software. The conceptual design is easy. Writing the 2 million lines of code is hard. Copyright works well for that.

If all you had is a reference document on the instruction set, are you saying the hard part is already done? Or that the hard part would be figuring out how to implement that?

But MIPS doesn't sell the instruction set reference. They sell CPU cores. So you have a blueprint that shows you exactly how the CPU works. The hard part is coming up with this blueprint, including things like designing caches to minimize power or structuring execution units to improve timing characteristics. That takes lots of expensive engineer time, iteration, testing, etc. But once all that information is distilled into a blueprint, it's relatively easy to actually implement it. Copyright protects you from literal copying of the blueprint. It doesn't protect you from someone just looking at your blueprint and implementing it.

I used to work on military radios, and we spent millions of dollars and years developing our technology. The bulk of that investment was coming up with designs that worked well. We shipped customers reference code. Copyright kept the customer from copying the code itself, but didn't stop them from looking at how our code worked and writing their own version. But the code was the easy part--once you knew exactly what the code should do, it was a matter of 3-6 months of fairly straightforward coding to reproduce it without literally copying the code. It wasn't like say Windows where the hard part is literally reproducing all those 10 million lines of code.

"It doesn't protect you from someone just looking at your blueprint and implementing it."

I don't think this is true. You are creating a derived work, after all.

Googling for blueprint and copyright shows that copyright disallows unauthorized implementation at least in the case of buildings.

MIPS also sells instruction set licenses if you want to build your own MIPS core from scratch.

That's a somewhat valid point, but protecting that sort of innovation with patents is still extremely heavy handed.

Just as software can be licensed with terms, surely these "blueprints" can be licensed with very specific regulations on how and where they can be used. Violating that agreement, which is very specific in nature, would be a lot more reasonable to enforce than what we have now.

Those patents that've been sold off are most likely going to be used to shake-down other companies and lock out innovation. The most exploitable will be cherry picked, shunted into shell companies, and in time some jury in East Texas will have to make a decision on how many billions in damages need to be awarded.

I think the key aspect you're missing, though, is that contracts cannot enforced against people who are not party to the contract. If some company sits down and reverse engineers your design, you have no recourse. If your design gets leaked, you have no recourse. You have no way to protect something which is complex and expensive to derive but simple to copy.

I mean neither of us have the data to show just how heavy handed or not the patent system actually is. I don't think it's particularly heavy handed, but then again my experience is in industries (defense, telecom) where "heavy" means something quite different.

At least outside the software realm, my intuition is that companies illegitimately ripping off another company's designs is a bigger problem than companies accidentally arriving at similar designs and patents being used to "shake them down."

I also think it's disingenuous to paint a picture where you have "100% innovation" on one side of the spectrum, with any patent rights as "locking out innovation." Again, because of my background I tend to see the innovation process as not a couple of guys in a garage doing some Kickstarter-funded web page, but companies working with universities and the government to build big technologies that require huge amounts of capital investment. The patent regime really facilitates those interactions, and I think increases innovation in doing so. When you're dealing with big organizations, a predictable set of property rights is much preferable to a fragile web of NDA's.

Just as an example, a really tight "intellectual property" regime would have to have...

This is a logical fallacy: for an act to be illegal, it does not follow that the state is empowered to preempt it from ever occurring at all, especially at the expense of basic civil rights. Conspiracy to commit murder is illegal; yet most police forces don't read every piece of communication to make sure it isn't a murder plan. Domestic violence is a crime; yet there are not spouse-monitoring CCTV cameras in every household.

OK, it's a logical fallacy, but you're splitting hairs. What's the point of "intellectual property" except to get some government to enforce it? "IP" laws don't empower individuals, in the way that say, free speech rights do. "IP" laws empower enforcers. So, yes, strictly speaking, I made a logical fallacy. In real life, that's the way "IP" is going to play out: strictly enforced.

What's the point of "real property" except to get some government to enforce it? That doesn't mean you need to enforce it as strictly as you can possibly imagine. By and large there is no government enforcement of patents--it's all left to private suits.

You can uniquely possess real property. That's actually what distinguishes "intellectual" from "real" property. The computer on which I type this is mine: I possess it solely. When I'm using it, nobody else can, there's simply no room o the keyboard. The same holds true for my other real property, my stove, my refridgerator, my toaster.

You can use my "intellectual property" at exactly the same time and pretty much in exactly the same way as I use it. The law about "IP" is not about property per se, it's about monopoly. Some government agrees to enforce your monopoly on some idea.

I'm not splitting hairs -- suing a company for cloning a CPU, versus spying on citizens to preemptively check that they're not pirating -- the difference isn't a hair, it's an enormous gulf. I'm not hostile to slippery-slope rhetoric, but you have to make it credible; this is over-the-top.

But the spying/surveillance is what we see happening: Google for "six strikes" and see what you find. Google for "The Unbearable Lightness of Monitoring", let me know what you find.

We also have a lot of "spying" precedents in the USA. Google "CALEA", "DCS-3000", "DCS-6000". Once something becomes illegal, almost nothing will be spared to suppress it. The NSA is almost certainly snooping on almost every US citizen, with the aid of telecomm companies. Google for "Room 641A". They aren't even looking for specific crimes, they're just monitoring.

Google for "crypto wars" to see what the USA wanted to do. Read especially the specifics of Skipjack that you can find. Ask yourself why cryptography was so nearly suppressed.

I don't really think this is over-the-top: it's almost factual, very little separates us from a really tight "IP" control regime.

Look, the "very ownership of ideas" rhetoric has a certain romantic appeal, but we allow people to own lots of things they didn't even create. The land under your house was there before you existed and will be there after you're dead, yet you're allowed to "own" it. We create property rights in various things because they are a useful way to structure social interaction. That's the only criteria for evaluation: usefulness. Romantic ideals have no place.

As for your hypothetical example, who says the strongest possible intellectual property regime is the best one, and how does that imply that no intellectual property regime is the right answer? We certainly don't aggressively enforce other kinds of property rights. You can't sue someone if their dogs crap runs onto your yard when it rains, even though in theory you could define that as an infringement of your property right. We try to balance things. Enforce rights enough to give structure to human interaction without enforcing too much and stifling that interaction.

I encourage you to look into the "moral rights" granted to creators under the Berne Convention, which the USA is a signatory (http://en.wikipedia.org/wiki/Moral_rights_%28copyright_law%2...). If "usefulness" of a category of law is the absolute criteria, then we should have less "IP". Plenty of math, economics, and research shows that "IP" actually causes less use, less creation, and more problems.

There are many examples of a division of labor between companies without intellectual property protection being involved. I have worked on many consulting projects where intellectual property protection of the results was not significant.

So for example MIPS could become a CPU design consulting company to fab companies that want to produce chips. It seems to me that this would reduce barriers to entry and improve allocation of resources.

Reply on news.ycombinator.com