Pure Awesome. Shame there's not a way to do the same thing to existing software patents. There are plenty (~40k/year according to the article) of bad, obvious nonsense patents already out there. It'd be nice if there were a simple process to appeal and invalidate them with similar demonstrations of prior art and obviousness.
I'm definitely looking forward to seeing some Wikipedia-level-OCD focused on this site to stop the roughly 100% of bad applications for new software patents dead in their tracks. Imagine a 2014 where zero new software patents were issued.
EDIT: Incidentally, patentlyobvious.com is just a parked domain at the moment. It seems like the obvious choice for a place to host a site like this.
Can you give an example of a patent being helpful in producing innovation and prosperity in the last 20 years?
IE, a patent that was an original idea, someone else tried to duplicate in a blatantly rip-off way before the patent expired, and the ensuing lawsuit lets the patent creator (who has their patent to-market in a useful state) take the copycat off the market?
You're assuming that patents promote innovation once they're defended in court.
Another argument would be that they promote innovation by (1) preventing blatant rip offs in the first place, (2) incentivizing individual inventors who invest their time and money with the assurance that their efforts won't be blatantly ripped off in the end by someone with more resources.
There are countless examples of truly innovative patents filed by individuals who devoted their lives to creating new things. Some of the inventors profited directly from their patents, others spent their life savings in court trying to defend their patents.
There are also countless patents for idiotic ideas, and countless idiotic patents that shouldn't have been filed in the first place. But saying all patents are bad might be throwing out the baby with the bathwater.
Absolutely correct. I used to work in this industry, and one of the 1st questions asked when we were thinking about testing a drug to treat a disease was "Does someone else 'own' this drug or do we?"
If someone else 'owned' the drug, then there was no point in pursuing it because the true "owner" would crush us if the drug ultimately proved useful in treating disease.
Interestingly, the people writing the drug patents use the same types of techniques mentioned by Joel...especially making claims as broad as possible.
The chemists that I worked with considered it an art form to be able to make broad claims that did not intersect with those of competing pharma companies.
There's a lot of bullshit that can go on w/ pharma patents, too, but the VCs would never fund a biotech startup if we couldn't defend "our" drugs.
While I'm sure a case could be made that IP protection is necessary in biotech, isn't the example you gave actually evidence of the the opposite -- that patents stifle innovation and research?
It depends on what you want to get out of the pharmaceutical industry.
I think it's the opposite. Since it costs so much $ to fund clinical trials and apply for regulatory approval, the lack of patents would stifle innovation because the pharma company that discovered the drug wouldn't be able to recoup their development costs if other companies were waiting in the wings, ready to sell a generic version of the same drug.
If development costs (especially regulatory approval costs) go down significantly, then I think you can argue that drug patents are an unnecessary evil. But if things stay as they are, then they are a necessary evil.
I think the idea is that then they couldn't charge exorbitant prices to recoup their investment costs. Drugs cost an enormous amount of money to develop.
Put yourself in their shoes. It's going to cost you $50MM-500MM just to test if a drug is effective and safe for treating a disease.
You sink all of that money in, with the hopes that your drug gets approved. And then, once you've done the heavy lifting, every generic manufacturer can come along and sell the drug because you don't have it patented?
You must consider though how the massive test run costs associated only exist because the FDA believed drug companies could foot the bill since they got monopoly distribution.
In a patent-free world, those regulations wouldn't hold weight. What would happen (I imagine) is independently sponsored r&d produces drugs for trial that the FDA itself puts through clinical trials on taxpayer money. You can't sell a drug without FDA approval still, but since drug companies aren't operating as super-for-profit businesses on chemical compounds to help people live, they would foot the bill to make sure its safe.
> Interestingly, the people writing the drug patents use the same types of techniques mentioned by Joel...especially making claims as broad as possible.
Systems (legal, computer, etc.) need to be developed with the full expectation that people will do everything in their power to exploit them.
I don't necessarily call that a good thing. It isn't a black and white pro that a business gets monopoly rights to distribute new drugs for 2 decades, even if most new drug research is funded for that profit.
I have to think that the patent in Spolsky's example cost at least a couple grand to produce/submit. I also would love to see the cost per rejected patent produced by the USPTO that comes from US tax payer dollars.
Actually you do have to worry about that. Since the USPTO is financed by patent application fees it has a perverse incentive to continue promiscuously granting as many patents as possible.
1) Pretty much each and every action has a fee attached [1].
2) Each rejection has a 3 - 6 month time limit to be responded to.
3) Most applicants will fight every rejection to preserve their rights.
4) Granted patents have (admittedly higher [3]) fees due only every 3, 7 and 11 years, whereas patents in prosecution typically have fees due every 3 - 6 months. However, granted patents are not guaranteed to be renewed, which means their contribution is reduced [4, 5].
So if you want to talk perverse incentives, the USPTO actually have greater incentive to issue more rejections and collect frequent short-term fees on applicant responses, than they do to issue patents and collect infrequent long-term fees on renewals.
This myopic myth-busting of yours assumes that the number of patent applications remains unchanged regardless of how many are granted when in reality more granted patents fuels the fire of more applications. If the USPTO did the right thing and severely limited or even eliminated software patents there's no question that the number of applications, and thus the USPTO's overall budget and staffing levels, would decrease dramatically.
Apologies for the late response. I don't check regularly.
1. There is no evidence to back your thesis that "more granted patents fuels the fire of more applications". See this graph [1] of allowance rates in 2000 - 2005 (for an explanation, ask a patent attorney or agent about "Dudas") and this table [2] for number of applications over that time period. There is zero correlation with the steep drop in allowance rates and the rate of new patent applications over the relevant time periods.
2. Why do you think your idea of "the right thing" is actually the right thing, if any such thing exists? Just because the HN echo chamber agrees with you? As I have shown time and time again, the Internet, especially tech media, has no real knowledge of how patents work.
3. The Supreme Court, let alone the CAFC, the PTO and various courts, cannot even define what "software patents" are. How would you begin to limit or eliminate them?
4. None of the above entities have found sufficient rationale to exclude business method patents, let alone software patents. Contrary to popular belief, it's not because they are stupid or technically illiterate, it's because they are very careful thinkers and make the best of what they can given their boundaries.
5. The budget and staffing of the USPTO is, unfortunately, mostly independent of the number of patent applications they receive. Currently, they generate more fees than they are allocated a budget for, and the federal government siphons away the rest.
1. Thanks for the second link - both "Total Patent Applications" and "Total Patent Grants" show a similar upward trajectory, both roughly doubling since the late nineties. The first link is uninteresting - acceptance rate fluctuations don't matter much in the face of steadily rising numbers of both applications and grants.
2.3.4. I come to my opinions as a veteran professional programmer, a named inventor on three US software patents (which I now regret signing off on), a software business owner, and a long-time avid follower of this issue. I'm sure that others, perhaps yourself included, may know more about the ins and outs of the patent legal system than I do, but I submit that you and your patent-defending ilk can't see the forest for the trees. The forest is that the vast majority of software-related patents are total bullshit. Given your role here as patent apologist, I don't think I will try to make a case to you why that is; I don't think that would be time well spent.
5. Find me a chart like your second link which shows the staffing levels of the USPTO over time. I'll bet it shows the same upward trend as patent applications and grants.
You may have to worry about the incentives, but you don't have to worry about any cost in tax dollars of rejecting applications.
Unless somehow they began rejecting so many applications, and those rejections all cost more than the fees collected from those applications, so as to total more in losses than the surplus revenue of the rest of the PTO (all other patent and trademark activity, including renewals) over its costs.
Since the USPTO is financed by patent application fees it has a perverse incentive to continue promiscuously granting as many patents as possible.
I would hope that filers for patents that are rejected would still get charged a fee. If they aren't, they should be. After all, it still takes up a patent clerk's time.
I think the problem is that the proverbial small inventor is extremely rare, and if they are indeed small enough for us to take pity on them, then a patent isn't enough to protect them anyway.
Make the prior art fee scale up by number of patent applications submitted by that entity, then. First one's free. Then $100, then start growing at some small-ish exponent.
> Shame there's not a way to do the same thing to existing software patents.
What's stopping anyone from crowd sourcing a database of prior art for existing patents that would invalidate them if it came to it?
A valid defense against patent infringement is to show that the patent shouldn't have been issued in the first place, right? So a database of Ready Made Legal Defense sounds good.
Showing a patent clerk prior art is much easier than showing prior art in a courtroom.
For instance at the minimum you need to bring in an expert witness to explain the prior art. Then you have to contend with the fact that they will bring in their own expert witness to explain how your example is bogus and you both get to cross-examine. etc. etc.
THe key part is recent changes in the law have given much more weight to community/public commenting on patents, so where these may have failed in the past it is possible for them to have more success today.
Further the patent climate has become more hostile towards bad patents in recent years, more so than at in point in modern history, this will also enable more possibility for success
Your assertion remains to be seen.
There are still significant downsides to submitting prior art to the PTO that also existed when bountyquest was around.
The main serious issues are still the same, in fact:
1. If you submit prior art pre-grant, that's basically that. You don't get to argue about it, only the examiner does. If the patent issues, you've now lost a possibly valuable reference if the examiner didn't use it well enough.
2. If you challenge in any of the post-grant proceedings, you'll be estopped from suing on anything you raised or could have raised.
There was an extremely depressing AMA on reddit a while ago where a patent examiner explained that what all of us consider "prior art" will be completely ignored by a patent examiner:
The bar for prior art is very high - it has to be published in a recognized medium. Most specifically, unless it has a date that the patent officer can verify and cite (and a self stated date on a web site is not "verifiable"), it can't be considered because it is not possible say for sure it came earlier than the patent filing. So - some random thing on the internet - not published. Even an actual real product made and sold by a company - not published. Even standard industry practise, established for years, if not written up and "published" somewhere, may not qualify as prior art. In one comment he says:
"You may be right, that is how everyone does it. But if there is no documented prior art for us to search, we are out of luck rejecting it."
This explains why so many things that software developers routinely do end up in patents. Some of them are just so obvious that publishing it in a formal way is redundant. Yet that is the same bar that the USPTO is applying for rejecting patents. So the patent system itself is enriching the pool of obvious patents that get through.
Anyone using this Ask Patents site really needs to sit through a mini-tutorial explaining these things before they start, or they will waste more time than they save.
What about the Internet Archive's Wayback Machine?[0]
And how can making a physical product not count -- surely any physical product sold in the United States has some kind of qualifying documentation associated with it which counts as publishing, even a manual?
The wayback bachine is used routinely, at least in Australia, to show that a web document was publicly available before the priority date of an application. One problem is that if the document happens to be from the domain of the applicant (more common than you'd think, some applicants are dumb like that), they can have it removed by archive.org. This is why examiners will make copies of the page from a date before the priority date when using documents accessed through the wayback machine.
> The bar for prior art is very high ... Even an actual real product made and sold by a company - not published. Even standard industry practise, established for years, if not written up and "published" somewhere, may not qualify as prior art
If that's true,[0] seems to me that the US has a much narrower definition of what can be part of the state of the art than Europe does. Here, the state of the art is "everything made available to the public by means of a written or oral description, by use, or in any other way"[1]
E.g. in the UK there was a famous case where a patent for a windsurfing board was invalidated because it was anticipated by a primitive board hacked together by a 12-year-old boy a decade earlier, and used by him on summer weekends.[2] That "the user was open and visible to anyone in the vicinity of the caravan site where the family stayed" was enough to qualify it as prior art.
[0] To be clear: as a Brit I have no idea about US patent law, and don't know whether the parent description of it is accurate. (It's not impossible that the USPTO's rules on what they'll look at when considering an application are narrower than what a court can consider prior art, just for practical reasons - a court can call and cross-examine witnesses to give evidence of oral disclosures or prior use, unlike a patent examiner. That's just a guess, though).
Yes, to be clear, (and I realize now that I wasn't above), what I'm talking about above is what a patent examiner is allowed to use to routinely disqualify a patent up front as part of the application process prior to granting it.
The examiner is not allowed to go on research expeditions to interview witnesses, or to locate, purchase and reverse engineer old products to prove they are implemented in such a way as to invalidate the patent. A court (or the litigants) can certainly do all that, but here we are focusing on what can invalidate a patent prior to it being granted. And for that we are limited to certain classes of readily verifiable types of evidence.
> “Sometimes you have a picture that you want to scale to different resolutions. When this happens, you might want to have multiple versions of the image available at different resolutions, so you can pick the one that’s closest and scale that.”
I read that Lance Williams invented this technique in 1983, but I remember seeing a documentary on Tron where they had to come up with a technique during the lightcycle race to reduce the Moire effect on the grid and I could have swore that they invented mipmapping to do that, even though it was a year earlier.
The patent trolls are already on it, patenting crowd-sourcing as a method of patent defense. This will inevitably result in the shut down of Ask Patents in a stunning fit of irony.
> Since patent examiners rely so much on keyword searches, when you submit your application, if you can change some of the keywords in your patent to be different than the words used everywhere else, you might get your patent through even when there’s blatant prior art, because by using weird, made-up words for things, you've made that prior art harder to find.
Wouldn't it also make sense to build up, maybe at the same time, a sort of "counter-thesaurus"?
As in: If you find prior art where the thesaurus method has been used to obscure terms in the patent, enter those as an example into a database. When another patent is looked up, individual terms that show up in the database have a "there are alternative terms for this" marker applied to them.
This might also make it a lot easier to make automatic search for prior art feasible again.
Does anyone know whether participating in a public forum like this could used in future proceedings against the user or their employer as evidence of 'willful infringement'?
I vaguely recall something about large corporations discouraging engineers from reading potentially relevant patents due to the possibility of greater damages if they were later proved infringing, since demonstrating that they didn't take any inspiration or details from the patent is quite difficult.
Not wanting to spread FUD or anything, but it strikes me as a potential exploit for patent trolls to discourage participation if it's a significant risk.
I have been asked by my employer (a large software company) to avoid looking at patents in any form. It is precisely because, as you cite, it exposes the entire company to willful infringement claims around any patents that I were to view.
I had a bad attitude dealing with patents in my last job, refusing to take part on the basis that I hadn't come up with anything novel. I noticed a member of staff with views on patents in line with mine taking a more active role. He took part in the patent meetings, but instead of merely offering nothing up, took active part in finding prior art, thus preventing time wasting patents going to the office. He took special pleasure in finding prior art by current employees at the same company, preferably in the same arm.
I am impressed by Spolsky's positive attitude to do something about the problem, I bid ask patents continued good fortune!
So, this sounds great and all, but I don't see any proof that it was ask patents that caused the examiner to find this, rather than the examiner's standard search?
In fact, the search history, on PAIR (look for 4-11-2013 SRFW Search information including classification, databases and other search related notes) does not say ask patents was used, nor does the search strategy (document code SRNT).
It could be the USPTO has not gotten around to noting this yet, but the only entry I see that could be related is the NPL entry, which of course, has no image available (god i love PAIR), and does not say it came from outside the search.
Don't get me wrong, I think askpatents is great, but i'm skeptical considering how examiners actually work.
The USPTO explicitly asked us to set up this site to help get the public involved in getting prior art to the examiners. (Press release:http://www.uspto.gov/news/pr/2012/12-60.jsp.) So, while it's entirely possible that the examiner might have eventually found this without us, which is far from certain, it seems reasonable to assume that they saw it on the site. Which means that even if you assume they would have found it without help - again, questionable, given the number of weak patents out there - at a minimum, we probably saved them a lot of time, which frees up more resources to look for prior art on the next crappy patent. And that reduces the odds that the clock runs out without them finding any, which leads to a lousy issuance.
I'm aware the USPTO asked you to set up ask patents. The USPTO has a number of pilot programs, and i'm involved in some.
My only question is whether AskPatents was how it got to the USPTO.
I don't see why it's reasonable to assume that they saw it on the site considering most examiners still don't do that sort of thing. Most of them search the USPTO approved databases, internally, and go about their way. This is in fact, what the search strategy/et al says happened.
Given that, i think the burden falls on you to show some likelihood here.
For example, do you have anything to say that any significant percentage of examiners (IE > 10%) use askpatents on a regular basis? I would be pleasantly shocked if you did.
The rest is not the argument i am making, the only argument I am making is: "Do you have any evidence that your site is the reason the USPTO found this?"
From what I can tell, your answer is "no".
If you do, great!
No offense meant, of course, my problem is if you declare victory when you were not actually involved, this will actually make things worse - people will think they are helping solve the issue, yet, if the PTO actually isn't using your data heavily, they aren't.
Again, don't get me wrong, i think askpatents is a great idea, and i'd hope and love to find out it's actually being used heavily by the PTO. I can't find any stats on that, and given the history of prior pilot programs/etc, i'm very skeptical.
No offense taken, and thanks for the overall support!
As to:
> "Do you have any evidence that your site is the reason the USPTO found this?"
Well, the office has told us directly that examiners are reviewing the site. So, when they then reject an application, based primarily on art posted on the site, that was asked for and posted there just prior to the actual review, it makes me highly optimistic that it's not all a coincidence, although I suppose that's theoretically possible.
Is it the examiners who post some of the requests for prior art? If so, it would be nice if their accounts or requests could be badged so other users can be reasonably sure that the research they do will be seen by the USPTO.
I like the concept of Ask Patents but at the same time I wonder if it is possible that participation on Ask Patents could be leveraged by malicious litigants to seek treble damages.
For example, if the company I work for is being sued by a troll for infringement, could the troll see that employees have been participating on Ask Patents and therefore a reasonable assumption could be made that they're reading/reviewing/participating in discussions on patents, and therefore they knowingly are infringing? Would this something that either I or my employer should be worried about?
From my brief experience, you're probably better off assuming that the default price will be triple damages, and any reduction from this is a lucky discount.
This is an excellent result. I worried when the idea of helping the PTO uncover prior art would be controlled by the attorneys filing the patent (which is to say they would be a filter between the examiner and the external sources) but if the examiners are going directly to the source then this will really put a crimp in bad patents being issued.
This is an awesome project. I've signed up and will see what I can do to help shoot down patents!
There is one true and important point in the article that isn't supported by the example given:
> This patent was, typically, obfuscated, and it used terms like “pixel density” for something that every other programmer in the world would call “resolution,” either accidentally (because Microsoft’s lawyers were not programmers), or, more likely, because the obfuscation makes it that much harder to search.
The patent uses "pixel density" to refer to the physical size of the pixels on a display. This is a fairly common term, with over five million results in a Google search. I've used the same phrase myself for over 10 years with the same meaning.
"Pixels per centimeter (ppcm), Pixels per inch (PPI) or pixel density is a measurement of the resolution of devices in various contexts: typically computer displays, image scanners, and digital camera image sensors."
Of course here we can see where the terms get a bit confusing: "...pixel density is a measurement of the resolution..."
But display resolution these days usually refers to the number of pixels, not their physical size:
"The display resolution of a digital television, computer monitor or display device is the number of distinct pixels in each dimension that can be displayed."
Advertisements and spec sheets almost always use "resolution" this way, for example the MacBook Pro specs on Apple's site:
"Supported resolutions: 1440 by 900 (native), 1280 by 800, 1152 by 720, ..."
Forgive me if this seems like nitpicking, and I completely agree with Joel's point here: patents often do use unusual terminology to obfuscate what they're talking about.
This just isn't a case of that. The patent is using the correct term, and it even does a very good job of explaining what it means:
> A particular characteristic of display components that may affect presentations rendered thereupon is the pixel density of the display component, such as a pixels-per-centimeter measurement. It may be appreciated that such characteristics may be independent of the size of the display component (e.g., two display components of the same size may present different pixel densities; conversely, two display components of different sizes may present the same pixel density) and/or the pixel dimensions of the display component (e.g., two display components displaying a presentation with a particular pixel dimensions may do so with different physical sizes).
> There is one true and important point in the article that isn't supported by the example given
I can't actually vouch for the "true and important" part. I must confess I put that in there to try to be nicer about what I was saying. :-) And it certainly seemed plausible enough that patents would be deliberately written in obfuscated language, I just can't say it's true or untrue from my own knowledge. I know pixels better than patents.
There's a great comment here that sheds some light on this:
One thing I don't understand. Suppose a patent examiner does not ask on Ask Patents about a patent application, but I would nevertheless like to submit prior art.
The Ask Patents site was created in partnership with the USPTO; there is an expectation that examiners will search the site while researching a pending application.
Trouble is, what killed the patent was prior disclosure from the researchers.
The same thing has happened to friends of mine; an undergraduate's summer research presentation may have betrayed patentable inventions. An effect of the 1-year prior art rule is to force researchers to keep mum about what they're doing and to generate greater numbers of incremental patents.
If you're interested in the free flow of information, using a researcher's own publications to kill patents may not help the cause.
Prior art from other work in the past? Bring it on!
I didn't think from reading it that the two groups at MS were the same - I assumed they were independent parts of the org, in which case it was fairly clear that this was an idea that various people have had without direct communication, and thus throws serious doubt on "non-obvious". I'm pretty sure that you could find plenty of other people who've had the idea.
Companies like Akamai used to (probably still do) have image resizing proxies which took something like a multi-resolution TIFF file as part of their extended CDN features. I can't remember when I first saw that in action, but I'm sure it was pre-2011.
I'm also not sure why using a researchers own publications to kill a patent would be bad in your eyes. If you think something is really, truly patentable you'd keep it under wraps until you did so, as far as possible. If you only decided later, once you'd published research and people were productizing that research - and you then patented and sued them - well, that's exactly the kind of thing that should be stomped on!
If you think something is really, truly patentable you'd keep it under wraps until you did so, as far as possible.
What this encourages though is, "If you think something is maybe patentable, you can't ever tell anyone about it until the invention is complete and fully functional." Nifty side projects occasionally become a big deal. For precision hardware, development time is often 5-10 years. For young researchers, it's critical to be able to talk about whatever it is that you do.
Completely agreed that bait-and-switch is unethical and wrong.
I'm not sure I understand what you mean by "prior disclosure" here. They filed a patent application: those are supposed to be public. The prior art Joel used was the documentation of the Win32 API from 5 years previously (and I'm pretty sure we've been using variable-resolution apps since earlier than 2008, and not just in Windows -- this was just an example). Could you clarify what the problem is?
App icons in OS X are interpolated between about 4 different sized images. I remember reading about this in the developer docs back in about 2002. I'd be surprised if Apple don't have pretty much the same patent on this already but registered 10 years earlier :-)
Mac OS has done this since much earlier. The classic Mac OS needed at least two different representations of the same icon, one for the desktop (32x32) and one for the application switcher (16x16). The application switcher, initially called MultiFinder, was introduced in 1987[1]. If memory serves, the various sizes of an icon were stored together in the ICNS resource of the app binary's resource fork. There was a developer tool provided by Apple called ResEdit which allowed you to view or change the icons of any application, in a nice GUI editor (of course).
I'm not 100% sure what you are confused about, but let me take a stab at it. Prior disclosure is public disclosure of the invention prior to the patent application. The one year rule states that once you publicly disclose an invention, you have one year from that date to file a patent application. Otherwise, your invention is considered public domain and cannot be patented. Does that clarify?
The problem I think you are referring to is that researchers run the risk of shooting themselves in the foot by prematurely presenting their findings unless they file an application in the process. Doing so would ruin their chances of being granted a patent.
I understand how prior disclosure works, my confusion is that nothing like that seems to have happened here. The patent was invalidated on the basis of documentation of a different piece of software with the same features released 5 years prior (albeit by the same company).
However, having read the ruling, I see it explicitly cites the prior-disclosure rule. So the bad patent system has invalidated a bad patent, but in a bad way.
The 1-year grace period is there specifically so researchers can publish and patent. What counts are the publication and filing dates. File within the year and you will be in a position to earn protection.
However, the 1-year grace period isn't available everywhere. The USA has it, for instance, but the EU doesn't.
An idea that might help in killing patents: Have one daily sticky post on HN (ala job post) with a short description of a pending patent and a link on where to post prior art. It is very likely some reader will be able to provide this prior art.
At 40,000 software patents per year, it would require 110 posted patents per day to actually stop them, or a significant fraction of that to make a serious dent.
One a day would be less than a 1% reduction in software patents. We need to shut them ALL down.
> How cool would it be if Apple, Samsung, Oracle and Google got into a Mexican Standoff on Ask Patents? If each of those companies had three or four engineers dedicating a few hours every day to picking off their competitors’ applications, the number of granted patents to those companies would grind to a halt.
I'm kind of surprised this isn't happening already.
In another patent thread some time ago I identified what I think is the core problem with software patents, which is abstractability. I used the example of a tractor being generalised into transportation, so I was nodding along with Joel's example.
This is because that's how software development often proceeds. We start with the concrete problem, then notice a pattern that encompasses a class of concrete problems, then a pattern that describes a group of classes of problems and so on. Building abstractions is literally what we do as a profession.
Now, as Joel points out, the rational strategy is to take the highest-level, most abstract version of your invention to the patent office to see what will get passed in. So patent applications are written like matrioshka dolls, with a super general case on the outside, and progressively more concrete descriptions as you go deeper. Somewhere near the bottom is the original thing that started the ball rolling.
It would be cool if there was a way to penalize high number and high ratio of rejected patents from a single company. OTOH I don't see a way that allows genuine mistakes from small shops while being immune from big bad corps acting via shell companies.
That is an excellent idea. (To clarify, you're saying a company would need to have both a high number and a high ratio rejected to be penalized.)
More generally, it seems logical to have the size of the penalty increase at a quadratic or even exponential rate, based on the number of rejected patents within a certain trailing time period[1], and scaled by the rejected/submissions ratio. So every rejection would carry a penalty, but the more that were rejected (both in total and as a percentage), the larger the penalty would be, at an increasing rate.
It might also be an idea to do this with individual claims as well as, or even instead of, entire patents, to also deal with the overly broad claims problem.
[1] Or even for all time, but with a weighting factor based on recency. Wouldn't want to get overly complex though.
This is great! I don't truly believe that no software patent should ever be granted, but we should definitely weed out the bullshit ones.
I would think mipmapping would be the core prior art, and that's from 1983. But I suppose anything over a year older then the application is good enough.
Also note there is another technique for patent manipulation which is to provide a provisional patent application (to start the clock) which can't be granted, then continually refine and the application as time goes by. In a worse case scenario you could taller a general patent into a specific patent based on someone else work, and have a patent already in place that will predate the new invention.
somewhat active on Ask Patents. In fact, I've submitted an answer that is pretty sure to kill at least one Google patent application, and possibly another from Uniloc.
A few comments on this article:
1) This is a very unusual case; most answers (and almost all questions) from "lay engineers" completely misunderstand the scope of the patent, since they don't even know what claims are. And even if they do, they are very lax at interpreting claims. And even then, most posters frequently misunderstand the terms used (case in point, Spolsky's very post! https://news.ycombinator.com/item?id=6084884). All this leads them to post irrelevant prior art.
That does not mean there are no useful answers at all; there are, but they mostly come from people who are somewhat versed in patent law (such as agents, lawyers and examiners). Some re-wording of claims, such as what Micah Seigel does in his posts, helps, but for the proportion of useful answers to go up, we need more education about how patents work for this to be useful. It's really not that hard; heck I did it!
2) It's wayyy too soon for a victory lap because that was only the first non-final rejection, for which a response has already been filed. Statistically, this application will undergo 2.5 more rejections [1] and (based on my guestimate) at least one Request for Continued Examination (RCE) before being abandoned or (more likely considering the applicant) issued with much narrower claims.
3) Patents are worded so not (primarily) to be obfuscating, but rather because of legal, technical and some silly historical reasons. For instance, pronouns are very rarely used because any indefiniteness can be cause for invalidation. Obfuscation will not help much, because you are not trying to get it past lay engineers, but patent examiners, who have a technical background and are (usually) adept at reading patentese. Complaining about how hard it is to read patents is like a Blub programmer complaining about Lisp. You simply need to learn the language to appreciate what you are reading.
4) Most "software" patents (which can't even be cleanly categorized as such) are not crappy, at least with respect to all other patents. There are studies presenting this view [2, 3], but it's also based on my experience having read hundreds of patents. Almost none are revolutionary, but just as few are really as bad as the media portrays. The PTO has gotten pretty good at finding prior art (interestingly around the same time Google came around), and the really broad patents are dying out.
The "crappy software patents" view is common mostly because tech media routinely publishes uninformed (or disinformed? [4]) rhetoric, mostly because they garner some easy rageviews, and audiences accept it without critical thought. I do think the bar for non-obviousness should be different, but solving that is a difficult, almost-philosophical problem.
5) In response to various comments on this thread regarding pay-for-prior art schemes, initiatives such as Article One Partners already exist.
I am not a patent lawyer or an agent, but I believe in the patent system, as I have actually worked for the mythical small-guy firm that was ripped off by the big guys and almost died, but eventually prevailed with patents. You don't hear these stories much because typically the small guys don't have the PR budget for it [4]. (And also because many of those with patents turn to trolls, who like to keep a low profile.)
I have only recently become personally invested in the patent system, but I want all inventions, including mine, to be truly novel and worthwhile. And I want people to get off their butts and do something rather than complain about patents on HN. This is why I support Ask Patents.
Re (2): In 1924, legendary federal judge Learned Hand [0] remarked that "the antlike persistency [sic] of [patent] solicitors has overcome, and I suppose will continue to overcome, the patience of examiners, and there is apparently always but one outcome."
Agreed. It is an unfair match. It is, statistically, almost a tautology: The set of examiners is necessarily limited by government funding constraints, but the number of patent attorneys and agents they contend with on a daily basis is only limited by the market, which is huge. The examiners will eventually be overwhelmed.
I think this is one reason that most examiners (at least IME) have their default mindset to "Reject! Reject! Reject!" Also, this is why something like Ask Patents is invaluable to even the odds.
I don't agree with the quantitative approach, but I can't help think that technology can help. Google has already (in my opinion) helped the PTO greatly narrow claims the past decade; similar technology can help even more.
I have some background in NLP. And I know it's surprisingly effective when it comes to domains with specific jargon (cf. Watson and medical language). I've lurked long enough to know some here (such as VanL) have already experimented in this area. Personally, I have toyed with the idea of constructing parse trees out of multiple technical texts and claims, "normalizing" them using ontologies, and trying to find matches (i.e. prior art) using various tree-matching algorithms. I have a feeling it would be very effective. (Maybe Google already does this!)
But that does not address the problem of identifying patents that are quantitatively invalid but qualitatively valuable. To me, that is the more important long-term problem.
This weighs against individuals too heavily. The UK for a time had a zero filing fee but they've gone back to a small fee again (to avoid getting so much chaff).
It's the renewal fees where you should be charging highly IMO; hyperbolic year-on-year increases would be an interesting option to model.
I know this sort of thing is unpopular in the United States, but how about a pay-according-to-your-means model? So individual inventors working for themselves could file quite cheaply, but a publicly held corporation with a >$1b market cap pays a much bigger fee for the same thing.
Or perhaps even a pay-based-on-number-of-filings model. First application is $500, with a doubling for each filing, up to a max of $10k. (Or whatever multiplier and ceiling you want to have.) For small filers, the legal fees will dwarf the USPTO fees; for large filers with dedicated legal assets, the USPTO fees are still fairly minor but might start to be large enough to deter some of the frivolous filings.
How much does it cost to set up a shell corp, $50?
>For small filers, the legal fees will dwarf the USPTO fees //
You can self file; though it's not generally advisable. $10k isn't even a blip for someone like HP who (at least in the past) markets themselves on the number of patents they have.
>4) Most "software" patents (which can't even be cleanly categorized as such) are not crappy, at least with respect to all other patents. There are studies presenting this view [2, 3], but it's also based on my experience having read hundreds of patents. Almost none are revolutionary, but just as few are really as bad as the media portrays. The PTO has gotten pretty good at finding prior art (interestingly around the same time Google came around), and the really broad patents are dying out.
That's only a person assessment, not a fact, as you are making out it is. The consensus in the tech industry among almost everyone (from software developers to tech legends) is that the patent system is broken and continues to routinely pass invalid and valueless patents.
>The "crappy software patents" view is common mostly because tech media routinely publishes uninformed (or disinformed? [4]) rhetoric, mostly because they garner some easy rageviews, and audiences accept it without critical thought.
No, it's because it's actually the truth and the software community is not nearly as susceptible to junk journalism as you make out.
>I am not a patent lawyer or an agent, but I believe in the patent system, as I have actually worked for the mythical small-guy firm that was ripped off by the big guys and almost died, but eventually prevailed with patents.
For every example of a situation as you described, there are 10 of a patent troll shutting down another small-guy firm or making them work for slave wages.
The pompous aspect comes (IMO) from primarily historical reasons. However, in my experience, the judgement of them being "utter crap" is simultaneously (1) mostly uninformed and (2) yet mostly correct.
Let me clarify: people think they are crap for the wrong reasons, such as vastly overestimating the scope of the claims. But they are (IMO) still mostly crap because they are ridiculously narrow and not necessarily clever... Yet they are still valid because that's what the patent system is tuned to.
I have said this before in other comments -- the way patents are examined is mostly quantitative: find enough pieces of prior art where enough snippets of texts cover the text of the claims, and you have a rejection. No thought is given to the quality of the claims. For instance, I'd wager the RSA patent (which even the most die-hard anti-patent folks typically admit is novel) would not be granted today because each aspect of the claims is mentioned somewhere in the prior art. It is the combination of steps which is novel. Yet todays examination process would shoehorn the claims into a ridiculously narrow set of steps that (the attorneys hope) would still typically be infringed.
Examination these days is a mechanical process whereby examiners just try to find enough references with language in them to cover all steps in a claim to issue an easy rejection. (Yes, the "rubber stamp" is more likely used to reject rather than allow.)
So how do you try to qualitatively judge a patent when you have only a few hours appreciate all the prior art out there and appreciate what is being claimed and to do so?
The answer is, you can't; the only way to have a scalable process is to issue actions on a quantitative basis rather than a qualitative one.
As I said, it's a difficult, almost philosophical question.
> The "crappy software patents" view is common mostly because tech media routinely publishes uninformed (or disinformed? [4]) rhetoric, mostly because they garner some easy rageviews, and audiences accept it without critical thought.
"Inventive patent filed covering obscure corner of programming" is not going to garner pageviews, for the same reason that "Grandmother walks home safe from 1000th time without being assaulted by ne'er-do-wells" doesn't appear on any newspaper outside The Onion.
Be aware that there is a backlog of "secret" patent applications in the US, where the patent application can be kept secret until such time as it is granted, at which point the easy invalidation described by Joel is not allowed.
For example the Twitter patent was granted, even though there is prior art (mine), and even though I submitted my prior art as soon as I found out about the patent. But I got a reply along the lines of "too late, the patent has been granted". It turned out that the application was kept secret until it was granted, thus guaranteeing protection from any prior art not already known to the patent office employee assigned to that patent.
The "new" US patent laws require an application to be made public after 18 months, but there could be thousands upon thousands of older patent applications benefiting from the secret option.
It would be great if others could attach bounties to certain patents. Of course then the patent examiner would have to pick "winners", answers that helped the most.
Also, like Quora, does the StackExchange system allow one to register one's fields of expertise, to have questions suggested to you? (I'm not a big participant on SE but I know it may be hiding that feature from me, because it slowly reveals features based on karma).
Anyway I'm sure this has been thought of before, just curious if it's on the todo list.
StackExchange has a "favorite tags" feature which will highlight specific tags on the front page (and for busier sites, it may even bias its display on them, though I'm not sure about that).
There is also an advanced "filters" functionality (visible at stackexchange.com/filters) which apparently lets you filter out all but certain tags, as well as have 15-minute/3-hour/daily emails with new questions in that tag. I have never used it, but it sounds like what you're aiming for.
You can post bounties on Stack Exchange if you have enough reputation to cover it. Whoever posts the bounty (not necessarily the patent examiner) gets to pick who wins the bounty. If they don't, it can be automatically awarded to the highest-scoring answer during the bounty period (one week). More details: http://meta.stackoverflow.com/questions/16065/how-does-the-b...
> How cool would it be if Apple, Samsung, Oracle and Google got into a Mexican Standoff on Ask Patents? If each of those companies had three or four engineers dedicating a few hours every day to picking off their competitors’ applications, the number of granted patents to those companies would grind to a halt.
Wouldn't the result of this be that the company who finds out that they have the earliest implementation of the code in question be able to get the patent for it?
The first publisher of an invention has a year to file a patent on it. Because of the length of time taken to get to a filing, by the time this prior-art shootout takes place more than a year will have gone by and the prior art will actually disallow anyone from patenting the invention, including the original inventor.
But that could easily be for an idea that's proven to be over 17 years old (<1996 and counting up). So they'd be digging through the floor, and might fall through.
Why is this acceptable? Would we allow police detectives to out source solving a crime or garbage men to outsource picking up the trash? Why can't the patent office just hire good people like this guy who can figure this stuff out in 10 minutes? Perhaps the cost of filing a patent should increase each time you do it so the same people don't just keep flooding the patent office with this junk.
If the patent office cannot do this work themselves, they should offload to the private sector: small bounties for prior work, paid out of a fraction of the filing fee. It costs a few hundred dollars to submit a patent... thus, they could easily afford to give out a hundred or so for each comprehensive dismantling.
TLDR: Give people financial incentive to destroy patents, and they will be ALL over it.
Its actually some part of governing delegated to the public. A part they can be good at. I wonder if the crowd can be authoritative, rather than just helpful. That would let them tackle obviousness, not just prior art I could see a stackexchange-like. It might be that they can tackle it now, if good methods/guidelines exist for objectively determining obviousness.
1. Swap the burden of proof: couldn't much of these issues with software(!) patents be avoided just by forcing the "inventors" to check for prior arts themselves and prove (somehow) that they actually did do that?
2. Make them pay: What about a (huge) fine when the patent does not get approved eventually or when prior art was found during the approval process?
"The Mexican standoff bit would be neat. The AIA allows for third-party submissions directly into the patent application file. I am not sure that any of the larger companies would do this, but I think a devoted engineer in some of the smaller companies could use askpatents to stave off predatory patents."
Nice win but I have to say the bit that really got me excited was this
"My dream is that when big companies hear about how friggin’ easy it is to block a patent application, they’ll use Ask Patents to start messing with their competitors. How cool would it be if Apple, Samsung, Oracle and Google got into a Mexican Standoff on Ask Patents?"
> An example might help. Imagine a simple application with these three claims:
> 1. A method of transportation
> 2. The method of transportation in claim 1, wherein there is an engine connected to wheels
> 3. The method of transportation in claim 2, wherein the engine runs on water
> Notice that claim 2 mentions claim 1, and narrows it... in other words, it claims a strict subset of things from claim 1.
> Now, suppose you invented the water-powered car. When you submit your patent, you might submit it this way even knowing that there’s prior art for “methods of transportation” and you can’t really claim all of them as your invention. The theory is that (a) hey, you might get lucky! and (b) even if you don’t get lucky and the first claim is rejected, the narrower claims will still stand.
I'm pretty sure this is not accurate. To my knowledge, the claims of a patent are considered only as a whole, not individually, so there should be no fear that this is somehow trying to patent "methods of transportation" broadly.
To make a broad patent claim, you can't have any narrowing claims on the same patent. Put another way, the maximum scope of claim by a particular patent is defined by the narrowest claim in the list.
Not a lawyer, but I'm pretty certain that you're wrong. There are 2 types of claims, dependent and independent. dependent are those that refer to a previous claim and expand on it, (2 and 3 above). Claim 1 above is independent. The dependent claims are taken together with those they reference, but the independent claims if accepted are taken independently.
This! What the author describes is exactly what happens.
Our lawyers even admit that. The company I work for gets sued for a bogus patent every week(!), so we were all ask to file patent applications. I didn't, but I got involved in some applications as co-inventor anyway.
This isn't - and not been for a long time - about inventions.
This is a great first step, but can we sustain this and keep more patents from being granted. Everything hinges on community involvement, so hopefully we can build and sustain a community that does this every day.
> The number of actually novel, non-obvious inventions in the software industry that maybe, in some universe, deserve a government-granted monopoly is, perhaps, two.
How about just increasing the costs for filing and maintaining a patent as a quick way of getting rid of some of the really weak ones? Say double the fees and see what happens.
Any flat increase would just make it less likely that the big guys will have competition in their "most patents filed" competition.
Also, given we just switch from first-invented to first-filed, it is particularly onerous for those garage-based small inventors as a big corp can easily justify the cost.
What might make more sense is to tax intellectual property. We tax physical property with much less value than a "successful" software patent (ahem, Eolas?), why not tax the non-physical property as well? If you make it progressive based on the number of patents owned/controlled, it will also limit the usability of maintaining warchests of patents for controlling markets.
Perfect. Tax them heavier as they age. That way the people trying to create usable products are barely hurt, but by the last few years it costs a lot to hold on to. And you can release your patent to the public domain to release it. That's more likely to hit trolls than innovators.
Sadly a patent examiner (in the UK at least) gets less than a day to do a prior-art search on the most complex applications in the most complex subject matter areas; half a day is probably the standard.
There are half-a-million patent applications per annum in the US alone (10 million since 1963). I'd estimate a global corpus of billions of patent applications. Prior art search extends though [in theory] to all publications.
You can see why prior art submissions are allowed from third parties. If you want to ensure only valid patents, or perhaps as few patents as possible, are granted then making such submissions is in your own interests.
Edit: Actually the gross figures are misleading, but remember these are complex 20+ page [in normal print size] documents on the whole. G06F11 - an IPC classification concerning error correction, detection and handling has ~9000 docs in the espacenet database for each of 2010/2011/2012.
I'm definitely looking forward to seeing some Wikipedia-level-OCD focused on this site to stop the roughly 100% of bad applications for new software patents dead in their tracks. Imagine a 2014 where zero new software patents were issued.
EDIT: Incidentally, patentlyobvious.com is just a parked domain at the moment. It seems like the obvious choice for a place to host a site like this.