His summary, paraphrased: first-file is as accurate as first-invent, but cheaper. In particular, it seems to be subject to less legal uncertainty.
Broadly speaking, the only reforms to patent law that really matter to startups are the ones that foreclose on patent trolling. Patents take longer to issue than the event horizon for most startups. In the 5+ years you'd wait to get one --- an award that merely offers the privilege of spending hundreds of thousands of dollars and many additional years to initiate lawsuits --- you'll already have proven your company or not.
I'm not a lawyer, though I once was studying to be one. In '92 I took a break from programming and went to law school (University of Washington) with the intent of becoming an IP, corporate, tax, and civil rights lawyer.
This was motivated both by curiosity about the law, and I was getting burned out on programming. We were basically a small contract programming shop, specializing in taking on firmware projects that were late, over budget, and failing and turning them around on a fixed bid and tight schedule. It was rather intense, and after four years of it, I needed a break.
Law school was a perfect extended vacation. While I was in law school, my former company had switched from the business of fixing people's messes to developing its own products and selling them, and I had done some programming on that part time and over the summers as I got over my programming burnout.
By the time I was finishing up law school, and only had left writing a paper for the intellectual property seminar I had taken, I hit a case of writer's block. I could just not get going on the topic I had chosen (intellectual property issues of usenet).
By that time the plan was to get the law degree, pass the bar, and go back to my old company as primarily the chief programmer and secondarily as in-house counsel on corporate and IP matters. I went back to work full time that summer and intended to write the paper when the writer's block passed and inspiration struck, however long that took. I never did get around to that, and that's where the matter stands today, 16 years later.
So, combine that with (1) a decent math and science background (BS Mathematics, Caltech, class of '82), (2) an above average ability to recall obscure but relevant snippets from magazines, books, blogs, and whatever other random sources I've read or seen, and (3) sufficient lack of a social life that I have time to Google obsessively before making statements of fact that I'm not sure of, and I can end up sounding like I actually know what I'm talking about!
Why doesn't someone start a project to crowd-source the reviewing of these patents?
There are 150k patents issued in the US every year, which works out as some 410 a day. For a small company, that'd be a gargantuan task. Let's say half of these are software patents. That's 200 a day.
All you need is about, what, 1400 programmers to volunteer a few hours every week to review the 1400 relevant patents and flag them up for "wait a minute, wtf". If those are then rejected, it will make spurious patent filing a costly and pointless endeavour. If the patent office isn't reviewing stuff, it sounds like we have to take that into our own hands.
1400 programmers for a few hours a week doesn't sound that impossible. Presumably the EFF could help on the legal side. Or am I missing something fundamental?
Because, not to put too fine a point on it, programmers suck at evaluating patents.
Patent language is pure, uncut crystal nerdbait. If you don't know how patent claims language works (and why would you?), most patents will appear ludicrous --- and that's assuming your programmer patent reviewers even know the difference between claims and background material.
Incidentally, reviewing patents prior to their issue isn't necessarily a win for defenders. I remember reading an analysis showing that defenders stand a better chance invalidating a patent in court than during the patent process. The crowdsourced review process could just end up helping patentholders by providing a first wave of claims strengthening.
It would be far more feasible to fight and win the battle to abolish software and business method patents.
1. Each claim of a patent is an AND statement, or if you like, an infringement checklist: If even one element in a given claim is missing from the accused product or process---and is not present by a "substantial equivalent" either---then that claim is not infringed.
2. In determining infringement, the different independent claims of a patent are OR'd together---only one of them need be infringed for the defendant to be liable.
3. A patent claim is invalid it if returns TRUE on the prior art, or on something that would have been "obvious" in view of the prior art at the time of the invention. How to determine obviousness in hindsight is perhaps the big question of patent law.
4. The claims of a patent are like a salvo of cruise missiles. The patent owner will launch as many claims as it can against a purported infringer. Some claims will miss, that is, they simply aren't infringed. Some claims will be knocked down by defensive fire, that is, by proof of invalidity. If even one claim gets through to the target, the patent owner is entitled to damages and (quite possibly) an injunction against further infringement.
What about an easier way for the public to file prior art? If there was a public database of patents under review and anybody could upload things that appear to be prior art to help the reviewers in the process?
I don't know that it's possible to abolish software patents. The Supreme Court tried, and were overruled by a lower court. Then those lower court precedents were invalidated, yet software patents are still not abolished. The US legal system is baffling.
The Supreme Court tried, and were overruled by a lower court.
what
The US legal system is baffling.
I can see why you'd think that. Could you try referencing which case you think involved a lower court overruling the Supreme court (like find a news story that mentions it or something)? That way someone might be able to clear up the misconception for you. I agree that civil procedure and matters of precedent can be very confusing, but I promise you the lower courts aren't going about overruling the Supreme Court, much as they might like to.
I promise you that if the legal reasoning in "In re. Alappat" and "State Street" were applied to Benson and Flook, those cases would have been decided the opposite way. The legal precedent in those Supreme Court cases was overruled by the Fed Circuit.
> The Supreme Court tried, and were overruled by a lower court.
Umm, no. Lower courts can not overrrule the US Supreme Court. (The Supreme Court can overrule on some issue and remand. The lower court may then rule as it did before based on different issues, but that's not the same as the lower court overruling the supremes.)
> Then those lower court precedents were invalidated, yet software patents are still not abolished.
Invalidated by whom? (And then there's the question of exactly what was done.)
> The US legal system is baffling.
It's a lot less confusing if you don't rely on the media to figure out what's going on.
The title "President Obama Could End Software Patents Today" pretty much guarantees that the "legal analysis" is bunk.
As to the "conclusion", In re. Bilski wouldn't invalidate Bezos One Click patent.
Also, one shouldn't assume that all software is a "mathematical algorithm".
And then there's the RSA patent - while one can't patent algorithms, one can patent using a specific algorithm for a specific purpose. (That's been done since before computers were invented.)
All you're saying about the title is that you disagree with it.
All you're saying about the conclusion is that you disagree with it.
One can't make statements about all software that will ever run on new technologies that we haven't seen yet. However, general purpose digital computers as we know them today are equivalent to Turing machines, so it is safe to assume that all software for them is mathematical algorithms. If you have some reason to question the consensus of theoretical computer scientists, please be specific about it.
My assertion is that what happens all the way from the Patent Office to the Federal Circuit is inconsistent with what the Supreme Court said. The RSA patent does not refute this assertion; it's an example of it. The RSA patent preempted all uses of a mathematical algorithm within a general-purpose digital computer. By the legal reasoning in Gottschalk v. Benson it should have been considered invalid.
> All you're saying about the title is that you disagree with it.
Anyone who writes that title is demonstrating significant ignorance about how the US legal system works. Any "legal analysis" by such an individual would have a broken clock's chance of being accurate.
And, the conclusion shows that you didn't get lucky.
> However, general purpose digital computers as we know them today are equivalent to Turing machines,
Yes.
>so it is safe to assume that all software for them is mathematical algorithms.
No.
I can describe any sequence of steps using mathematics. However, that doesn't make all such steps a mathematical algorithm.
If it did, then no sequence of steps would be patentable under the "no patenting of mathematical algorithms" rule.
Do you really want to argue that no sequence of steps is patentable? Or, do you think that some sequences can't be described mathematically?
> The RSA patent preempted all uses of a mathematical algorithm within a general-purpose digital computer.
No, it didn't. The RSA patent explicitly says that it's for a specific use of a given (class of) algorithms, not all uses of those algorithms.
Does your database include data on pending patent applications?
The database only includes data on Published Applications in accordance with the 18 month pre-grant publication rules. Pending patent applications where the applicat has elected to not publish prior to grant remain confidential.
Still though, I'd love to be a part of that kind of overhaul if there were some other way to get pending patent applications.
Oh, so the patents to search and crowd-review would be public. So crawl the search database and have those new results for review.
Looks like another group is doing this in conjunction with the USPTO http://peertopatent.org/. I guess you'd have to have some way of bringing the concerns to the USPTO for review, a relationship with them.
I wonder if it would be worthwhile to keep the patent review comments around, that way if someone found themselves in a lawsuit they would have a starting point for defense.
I thought the same thing, until someone pointed out that if you publish information on your invention, it makes it impossible for anyone else to file after that. (Prior art.) However, YOU can still file up to a year later. (In Europe, it works as prior art against yourself, though. Meaning you can't file, either.)
That effectively makes it 'first to publish' rather than 'first to file'.
I am not a lawyer, and neither was the person who told me this.
I believe under the new law you would need to show you employed your invention as part of a business prior to their filing in order to have a case for prior art.
I'm an entrepreneur and inventor. I want to make a record here that I have no fucking idea whether publishing my invention is enough to have Prior Art, or whether I need to state the use of the invention in my business case, or whether I need to file a patent.
As an entrepreneur, an "inventor", and a patentholder: I believe this is simple:
(a) if you're planning on staking a claim on a patent, get a patent lawyer (no part of this bill reduces the need for IP lawyers; whether you like the concept of requiring IP lawyers or not is besides the point).
(b) if you're not planning on getting a patent (and I think you shouldn't, for simple pragmatic reasons), just publish what you're doing as soon as you can, and you'll have established a prior art track record.
The reform we're talking about changes the mechanics of competing claims for a patent, but it doesn't make prior art harder to establish (my understanding is that in a variety of little ways it makes prior art easier to establish).
http://news.ycombinator.com/item?id=3011427
Much is being made out of "first-to-file" vs. "first-to-invent", but the name is scarier than the actual practice. 'tzs explained that one too:
http://news.ycombinator.com/item?id=2976041
His summary, paraphrased: first-file is as accurate as first-invent, but cheaper. In particular, it seems to be subject to less legal uncertainty.
Broadly speaking, the only reforms to patent law that really matter to startups are the ones that foreclose on patent trolling. Patents take longer to issue than the event horizon for most startups. In the 5+ years you'd wait to get one --- an award that merely offers the privilege of spending hundreds of thousands of dollars and many additional years to initiate lawsuits --- you'll already have proven your company or not.