Hacker News Reader

Comment by freehunter | Hacker News Reader

Especially in the tech world. If you're not first on the market with something, in many cases you might as well be last.

Problem is, this keeps out small companies trying to innovate beyond their corporate brethren. If I come up with something completely new and awesome, it might be two years before I can start manufacturing it and it might still be new and innovative then. HTC might have it out in 18 months. So during the time I'm designing and testing this equipment, I have to maintain complete secrecy through the entire supply chain, because if one link in the chain goes down, my billion dollar rivals have run off with the biggest idea of my five cent company.

Raising the bar and lowering the time frame would lower the barriers to entry moreso than removing patents altogether. Two years to get to market and then 6 months after release would be close to sufficient.

View on HN

a) You probably didn't come up with something new and awesome worth the court's protection (sorry)

b) If you did, HTC is going to use it anyway and bury your suit in the courts until you run out of money (sorry)

c) Either way, your example only applies to hardware or you wouldn't have a supply chain or manufacturing

Ultimately, theft of the biggest idea of your 5¢ company doesn't warrant tens of billions of dollars of legal overhead.

Perhaps I agree that raising the bar fixes the problem: if your invention can be implemented by a general purpose computer, it does not deserve patent protection. And that's "can", not "is" : if it is at all possible to implement an invention in software, it is essentially math and does not deserve patent protection. Patent protection is for mechanical inventions (wheat threshers) and material transformations (metal casting). If pharma deserves protection, it is the exception, not the rule.

Well, it depends. If I had a truly great idea and also had interest in forming a cell phone company, why would I settle for just selling the idea (and why would I attempt to sell it if during the negotiations, the idea could be stolen at any time with no protection for me?) when I could be making millions selling a phone with my great software innovation?

The point is, if there's absolutely no legal protection on ideas, ideas are worthless. People use the argument that math is universal and doesn't deserve protection, but the reality is that math is hard. If you come up with a math solution that is really clever, I feel you should be entitled to some kind of protection on it. What is hardware but applied math? Does programming not take man-hours of research and development? Does every line of code paid for make it to market? If we're arguing that R&D costs are what make sense for patents (like the pharma arguments I hear) then maybe you should have to show an itemized bill for your R&D and patent length could be adjusted so it's yours until you recoup that investment.

We trash on Edison for stealing ideas and beating the inventor to the PTO. The same thing could happen in software. The solution I see to this is making sure the original inventor has a reasonable time frame to capitalize on the idea before it becomes a free-for-all. By that point, the market has had a chance to speak and recognize your innovation. What we can all agree on is that patents shouldn't be a lifetime income source based on lawsuits.

> if it is at all possible to implement an invention in software, it is essentially math and does not deserve patent protection

The current patent laws are insanity, but I don't agree with this. The class of mathematical ideas too abstract to patent should be distinguished not just by whether or not an idea can be modeled mathematically (they all can!) but by how fundamental the ideas are. The goal should be to protect abstract ideas that will naturally be discovered independently by mathematicians. Complex software that solves some real world problem shouldn't be disqualified on these grounds, since it is almost of exactly the same nature as hardware that achieves the same goals.

Any idea can be modelled mathematically, but that doesn't meet my criteria. Modeling is an estimation of how something works, but it isn't the mechanism by which something works.

The current state of software patents results in people filing three sets of claims: the invention itself, a method that implements the invention, and a computer system that implements the invention. The rationale is that if invention claims (complicated math) are not enforceable, method claims might be, and if neither of those are, perhaps claims that mention hardware claim might be. You'll note that none of these motivations are strong, and that's because it has been left up to the courts to decide, and there is very little case law around software patents since everyone settles out of court. We think they are patentable because someone decades ago successfully enforced a patent regarding computer controlled curation of plastic.

Complex software that solves some real world problem shouldn't be disqualified on these grounds, since it is almost of exactly the same nature as hardware that achieves the same goals.

What makes complex software more valuable than simple software? Any engineer worth their salt will claim the precise opposite. Also, a product is not in itself patentable outside of the trade dress. You first have to find the novel invention inside the product, which is generally a simple idea.

Modeling is an estimation of how something works, but it isn't the mechanism by which something works.

Let's look at a concrete example. Suppose someone produces a device that contains some generic components: a centrifuge, spectrum analyzer, etc. and sells this device on the market. Someone else comes along and develops software that, with the aid of one of these generic devices, can detect a blood defect or airborne pathogens or whatever. Do you agree that this software should be patentable?

Maybe you'd say that the software isn't patentable, but using the software with the machine is? And attempting to sell a clone of the software wouldn't be patent infringement per se, but if it was clearly intended to be used with the devices would be inducing infringement and also a tort?

If that's your position then we agree in spirit, I think, though I think that looks mostly like double talk to me. If you don't think such a device should be patentable on the grounds that the only new component is software, then we disagree.

What makes complex software more valuable than simple software?

I didn't say it was more valuable, and I don't think complexity should even be a requirement. It's just that software that pertains to the real world tends to be complex.

Do you agree that this software should be patentable?

No.

Maybe you'd say that the software isn't patentable, but using the software with the machine is?

No.

If you design a better blender, it is an visible improvement that can be immediately adopted by existing manufacturers before you can capitalize on your invention. There is little to prevent this from happening. It is unlikely that you will create a better blender, but should you manage to do so the people of the United State of America will grant you exclusive production for a limited number of years.

If you create software that better controls mixing equipment, you can license your software to existing manufacturers. You are protected from competitors by copyright and trade secret. You can license the idea to companies by developing software directly for their hardware and only providing them object code. If they create a new product based on reverse engineering your object code (clearly for reasons other than interoperability), you can seek damages.

Patents are for inventions which are expensive, visible, and not otherwise protected. Software, like literary works and the formula for Coke-a-Cola, has other protections. When patents are applied to industries where invention is common, hidden, and independent invention is likely, the system falls apart.

You seem to have a very "Web 2.0" viewpoint. Yes, all the consumer-facing web startups pimped here on HN aren't really doing anything where patent protection would be beneficial. But in the embedded realm, there are a lot of situations where software might merit patent protection.

E.g. my wife is pregnant, and we recently got an ultrasound. The ultrasound machine gave us a surprisingly detailed three-dimensional image of the fetus. I'd be willing to bet that the algorithms for generating these visualizations were not something you could whip up in a few weeks using Rails. Without patent protection, the maker of that software has to get into the business of building ultrasound machines. He can't just sell the patent to someone who builds ultrasound machines. I think there is a lot of value in being able to facilitate such specialization of labor.

I'd wager that the algorithms used to produce that 3D ultrasound image are also widely used and documented in academic contexts for scientific imaging. The test for patentability isn't "beyond the comprehension of a Rails developer," it's (or ought to be) "beyond the comprehension of all experts in all related fields."

The test for patentability isn't "beyond the comprehension of" it's "is not obvious to." I used to work at a company that developed algorithms for wireless devices. They were comprehensible to experts in the field. But they certainly weren't obvious. If they had been, we wouldn't have spent years and millions of dollars coming up with them.

And if you spent years and millions of dollars coming up with it and some other company somewhere else was doing the same and in the end you had very similar if not identical algorithms why is it that one of you is entitled to use your own creation and the other is not?

That's the thing I can't get past with most of your arguments on here. If an algorithm is truly unique and worth the patent then in reality it shouldn't need the patent at all.

I also don't understand the "I invented an extremely complicated algorithm for solving X" mentality. We all stand on the shoulders of those who came before us in solving more and more complicated problems. Without the software, algorithms, and general knowledge of computing that have been made available to us we'd not be able to take that and build ever more complicated systems on top of that.

I think our differences in views stems from the types of patents we have in our heads. People on HN think of software patents as patenting an e-mail notification indicator, where independent invention is likely to be very common. My experience has been things like complex signal processing algorithms where the probability of independent invention (within the patent period) is very low.

Besides the probability of independent invention, the other key difference between the two cases is that in the former, you could never sell someone the idea of e-mail notifications in the absence of patent protection. In the latter case, the idea itself is very valuable--the patent just provides a legal structure that facilitates transactions.

As for the "standing on shoulders" angle, I think it's all about the difference between applied research and fundamental research. There isn't any money in giving away the fruits of your labor for the good of humanity (which is why top students from Harvard, MIT, etc, head to investment banks rather than fundamental research labs). And even amongst the altruisticly-minded, some fields are much more engendering of generous feelings than others. Nobody thinks the joy of blessing the world with more efficient filters for chemical plants is sufficient compensation in and of itself for the work put into designing those filters.

I'm trying to think through the way competition would work here.

Imagine I'm the guy building the ultrasound machine. Company A comes to me with this new algorithm they've developed to do nifty processing of the data. Things I care about:

  1. Does it work
  2. What kind of support do they offer
  3. Does it run on my hardware
  4. Will they continue to come out with updates and improvements
  5. Do they have the right medical approvals

In any piece of data analysis code, the vast, vast majority of the actual code will go into stuff only orthogonal to your nifty algorithm. And the thing is, the stuff I'm really paying for is that vast majority. I want it to run on my systems, and not crash, and have a nice UI and...

Even if Company B comes along with software that uses the same algorithm, or perhaps I decide to write my own software, I still need to rewrite this vast majority of code that is not the algorithm itself. And, as a hardware guy, I'll be bad at that. As Company A, you get to compete with Company B using that.

This is an ultrasound machine, not a tablet. You're going to get a reference implementation which you'll integrate (or rewrite completely) into your own software according to the exacting processes specified for medical devices. It's not like licensing Android from Google.

As for competition... I don't understand the dichotomy you're creating. That "vast amount" of simple code (the UI, etc), is protected by a different government-ranted monopoly: copyright. That's the only reason A can compete with B on that basis. But why is a large body of simple UI code worth copyright protection, while the small but complicated imaging algorithm isn't worth patent protection?

And your example doesn't address the specialization issue. Why do we want experts in ultrasonic imaging competing on UI and updates/support? Isn't it better for them to be able to focus on new imaging algorithms (their core competency), and sell those to someone who has an expertise in medical UIs?

I use this example a lot, but think about ARM. Aren't they a perfect example of the patent system working correctly? Isn't it great that they can sell their IP and let NVIDIA integrate it with a graphics core for consumer devices, etc, without forcing ARM to go into the business of manufacturing every type of microprocessor people might need?

It's not clear why ARM can't copyright their specific microprocessor designs and license based on that.

Because that only protects against literal copying. It doesn't prevent clean-room reimplementation of innovative circuits, structures, etc.

It also protects against derivative works.

I don't think it's obvious that they should naturally be allowed to prevent a clean-room reimplementation. The real innovative and creative work is in the designs that would be protected by copyright - the alleged "innovative circuits" subject to patent are really just being used as patent traps to prevent a clean-room reimplementation of the architecture - as anti-interoperability measures, in other words.

That's almost offensive - you seem to have a very ignorant viewpoint. Have you never heard of licensing? Research teams regularly build libraries that become incorporated into products that they do not sell. They can also license the math (which is not patentable) and have another team actually write the code. Why does the licensee bother to pay? Exclusivity, maintenance, further improvements, etc. Ideas that are not patentable can still be protected as trade secret, making them commercially valuable. If a company does not pay their licensing fees, the researchers might choose not to work with them in the future, providing a competitive advantage to other companies that do.

Since you don't show a firm grasp on patents, let me provide an example of the problem. Remember that patents are a legal protection ignorant of independent discovery. Say that you created a new, successful communications tool. People are often bothered when someone sends them a message while they are giving a presentation, so you change their status from Available to Busy when their calendar says they are in a meeting.

As you become a big commercial success, IBM sues you for infringement of their patent on changing chat availability status based on calendar free/busy. They seek damages and an injunction until the matter is resolved. They will settle for 150% of your company's worth, or they will pay their lawyers to tie this up in the courts until nobody cares about your product or you run out of money.

Hopefully it is obvious that patents do not help innovation in this situation. Some will say that this is a case of a bad patent, and the solution is just to ensure that patents are good. The problem is that only a couple of people in the world could tell the difference between a bad software patent and a good one (none of them are judges or lawyers), and who are they to rule on IBM's lawsuit against you?

It is now on your head to request re-examination and prove that someone else had already created a product that uses a personal calendar to change away status based solely on information in trade publications and product manuals (if you can find them). Alternatively you could attempt to prove that the invention was obvious, which requires you to recreate the environment in which the patent was filed and convince a non-technical examiner that an average programmer in that environment would have found it to be obvious.

But let's cut to the chase and say that you run out of money. IBM can litigate their toilet paper patent for centuries without noticing, and you probably have a couple months of runway. This is what it means to have patent protection.

> Have you never heard of licensing? Research teams regularly build libraries that become incorporated into products that they do not sell.

Licensing is based on copyright, a different government-granted monopoly. Why should software libraries be protected when software algorithms aren't? What you're saying is that you should be able to protect your "yet another JS framework" under copyright, but not something like a power control loop in a wireless device under patents.

> They can also license the math (which is not patentable) and have another team actually write the code.

You can license it, but you have no legal protection to keep people from copying it.

> Ideas that are not patentable can still be protected as trade secret, making them commercially valuable.

1) you can't transact in trade secrets; 2) trade secrets are of no use when embedded as software in devices.

Say I come up with a new power control loop for wireless devices. The loop allows radios to cause less interference while maintaining bandwidth. Say I keep it as a trade secret. If I put it in a radio and sell it, someone can reverse-engineer the firmware and copy the design of the power control loop. Trade secrets won't protect me. Moreover, I have to get into the business of building every kind of radio I might want to sell. I can't just sell the power control loop to someone who is good at building radios, because trade secret doesn't cover things you voluntarily disclose. Imagine the situation: I go up to Raytheon and say "I've got a great power control loop, I'll sell it to you for $50m." And they say "well, how does it work? We're not paying $50m without understanding what we're buying." And you say "well pay me first than I'll tell you!"

> Hopefully it is obvious that patents do not help innovation in this situation.

Agreed, but that's a phenomenally bad patent.

> Some will say that this is a case of a bad patent, and the solution is just to ensure that patents are good.

Yup.

> The problem is that only a couple of people in the world could tell the difference between a bad software patent and a good one (none of them are judges or lawyers), and who are they to rule on IBM's lawsuit against you?

There are hundreds of IP lawyers with strong technical backgrounds who could tell the difference between the patent in your example and say the OFDM patent: http://worldwide.espacenet.com/publicationDetails/biblio?CC=....

The "chat availability status" patents are getting a lot of press on HN, but I spent years working at tech companies that produced patents along the lines of the OFDM example linked-to above. I think a lot of people on HN don't have the exposure to embedded software to be able to really appreciate what the other side of the patent coin looks like.

Why should software libraries be protected when software algorithms aren't?

Independent invention.

> They can also license the math (which is not patentable) and have another team actually write the code.

You can license it, but you have no legal protection to keep people from copying it.

To keep other people from copying it. The people that you license it to will be contractually bound. Perhaps, but it will be expensive and time consuming for them to copy it. Patents are for a limited time, so this is an argument tied up in a comparison of how long it takes someone to unravel a complicated system, how long a patent lasts, and how much the inventor would license it for.

I think a lot of people on HN don't have the exposure to embedded software to be able to really appreciate what the other side of the patent coin looks like

That may be true. If you have made inventions such as these, you should protect them with trade secret because chat availability status patents are far more excessively damaging than strong signal processing patents are valuable.

> Independent invention.

Nobody is independently inventing the OFDM patent I linked to, at least not within the patent term. You're taking an edge case and making the debate all about that edge case. I think it's quite reasonable to have genuine independent invention as a defense to patent infringement, FWIW.

> You can write a contract for anything.

You can contract for anything, but you can't enforce the contract against anyone not in contractual privity.

> That may be true. If you have made inventions such as these, you should protect them with trade secret because chat availability status patents are far more excessively damaging than strong signal processing patents are valuable.

As I described above, trade secrets don't allow the crucial feature of being easy to transact in. And you have absolutely no basis for your claim that "chat availability" patents are more damaging than signal processing patents are valuable. You can point to the Apple litigation on one side, and I can point to ARM, Ltd. on the other, etc.

You're taking an edge case and making the debate all about that edge case.

Sadly, I see this the opposite way. Quantity-wise good, innovative architecture patents are in the vast minority. If we are to deal with software patents one at a time, the combined effort dealing with bad ones far outweighs good ones.

You can contract for anything, but you can't enforce the contract against anyone not in contractual privity.

Hah, I already updated my comment based on your updated comment! HN isn't the best for these kinds of discussions.

Nobody is independently inventing the OFDM patent I linked to, at least not within the patent term.

I have been thinking for some time that we need a collection of good software patents. Not having read the OFDM or ARM patents (for which I'm not qualified anyway), I don't know if they are "good". But I do know that one day we're going to invalidate a large percentage of software patents or watch innovation happen elsewhere. When that day comes, it will be important to have a list of things that are valuable, or else we'll end up with the bar that I described (anything computable is not patentable).

Unfortunately, I don't know of enough of them to seed such a database. Maybe fast square root or some of Carmack's graphics work. And as such I'm okay with seeing them 100% invalid. If you disagree, perhaps we can chat about it (email's in my profile).

Say I come up with a new power control loop for wireless devices. ... If I put it in a radio and sell it, someone can reverse-engineer the firmware and copy the design of the power control loop. Trade secrets won't protect me. ... Imagine the situation: I go up to Raytheon and say "I've got a great power control loop, I'll sell it to you for $50m." And they say "well, how does it work? We're not paying $50m without understanding what we're buying." And you say "well pay me first than I'll tell you!"

I think that this situation could be resolved with a careful NDA and later a license, but you've provided some compelling examples that I think are worth discussing. Perhaps the end of software patents comes with a new sort of hybrid of trade secret and license? The only way to keep things safe in the future appears to be keeping valuable things secret (ask any DRM provider). Even now, how would you enforce a patent against a competitor with a DoD contract, or abstractly with a military contract in another country? The invisibility of software complicates patent enforcement, but not trade secret or licensing.

There are hundreds of IP lawyers with strong technical backgrounds who could tell the difference between the patent in your example and say the OFDM patent [snip]

Okay, let's take that argument and run with it. If there are so very many talented IP lawyers with strong technical backgrounds, then why is patent law so bad in the first place? Why are we even having this discussion?

I think there are two reasons. First, I don't think there are enough good IP lawyers covering software patents and software patent law. Why? Software is a very new field and it is changing very quickly. Even the foundational structures of our industry are less than a century old. As such, it's very easy to sneak through "obvious" patents simply because they don't look obvious to someone outside the field. Second, the patent attorneys who do understand how software works and how software patents can be used to constrain and thwart software development all seem to be employed by patent trolls.

For example, let's look at the recent Kelora patent case [1]. In this case, Kelora patented "parametric search" (e.g. drill-down search). They then filed suit against basically every online retailer that allowed shoppers to drill down by category. Now, to a programmer or a web developer, the concept for drill-down hierarchical search is obvious, maybe even a bit blasé. But to someone who isn't as well versed in software development, it can definitely look like a new innovation that deserves patent protection.

Fortunately, in this case, the patent was overturned. But not before it did its damage. Before Kelora took on the big companies, it managed to win settlements from a number of smaller firms [2]. That highlights the true cost of software patents. The cost isn't mainly from the patents that are contested in court. The cost is mainly from the patents that aren't. It's the cost of a business that has to shut down because it violates a patent it didn't know existed. It's the cost of man years wasted because a product has to be redesigned to avoid patent litigation.

Yes, patents do protect innovation. However, they protect certain innovations at the cost of making other innovations much more difficult. In software, especially, they form a tragedy of the anti-commons, where, in effect, your competition has veto power over your innovation. It is increasingly clear to me that these costs do not justify the limited benefit that patent protection brings for software. That is why software patents should be abolished. They are simply not worth the cost.

[1] http://docs.justia.com/cases/federal/district-courts/califor...

[2] http://techcrunch.com/2012/05/26/kelora-obvious-software-pat...

[3] http://en.wikipedia.org/wiki/Tragedy_of_the_anti-commons

An algorithm can be protected as a trade secret. Should a clean-room implementation of an algorithm be prohibited?

http://en.wikipedia.org/wiki/Phoenix_Technologies#Cloning_th...

You're mixing up copyright, patents, and trade secrets.

The Phoenix BIOS case was a copyright infringement case--IBM wasn't claiming that there was anything about the "how it works" of the BIOS itself that was protected, but that Phoenix must have copied the actual code.

Should clean room implementations be allowed when the "how it works" is the interesting part? Say I spent years developing an ultrasound imaging algorithm. I sell an ultrasound machine with that algorithm. A competitor comes along and says "gee, these work really well, I wonder how they work?" So they reverse-engineer my design, and do a clean room implementation of it, instead of coming up with their own design.

It will take them a good deal of time and money to clean room reverse engineer a non-trivial algorithm. During that time you can profit from your invention, and advance your product. It will probably cost more money from them to clean room reverse-engineer your product than to just license it from you. Phoenix BIOS only happened because the code was small and the reward for interoperability so artificially large due to IBM's control. If the BIOS just did something useful (a) there would probably be academic papers on how it worked, or (b) other people would probably have just licensed it.

Congratulations to you and your family.

Ty!

Reply on news.ycombinator.com