Joi Ito's Web

Joi Ito's conversation with the living web.

I am glad that Europe has once again rejected software patents by voting 648 against and 14 for the ruling of the patent - software directive.

I hear that arguments have been made that software patents are helpful for innovation and that venture businesses may in some way benefit from software patents. I can of course imagine cases where software patents might be helpful for startup companies, but from my personal experience, they are generally more of a burden on innovation at the venture level than a benefit.

Generally speaking, filing for patents is an expensive and time consuming task. Most startup CEOs don't understand and can't afford a patent strategy. I have done a number of calculations on the cost of filing and maintaining software patents, and one estimate we did for a company that I am working on was that it would cost about $750,000 to file and maintain a single patent in the major markets over the lifetime of the patent. Most companies I invest in raise only $1M or less their first round. In addition, to properly protect a technology and continuing developments around a technology, a portfolio of patents must be filed or you can be "surrounded" by application patents and derivatives filed by competitors. In away, filing a patent is practically like putting up an ad balloon for people to see where you are focusing.

Some startup companies I have looked at and worked with have in fact, invested in a portfolio of patents, but from my experience, most of these companies end up spending so much time on their patents that often the products never make it to market. The patents just become fodder for some large company when they are purchased in the bankruptcy fire-sale.

For companies who are working in a patent riddled space, I definitely do a mental calculation of the added risk of litigation and subtract that value from the valuation of the company or decide to not invest at all. I've heard to software patents referred to as land-mines in this context. The problem is, big companies gobble up patent portfolios from bankrupt startups and then have teams of lawyers who use these to go after competition. There is a measurable chilling effect. (Note also that some of the technology oriented anti-file-sharing bills that have been proposed will have a similar effect.)

The only practical use of software patents that I have seen are defensive. Many Internet companies that I have worked with have one or a few broad software patents that they wield to threaten potential assailants. Typically, these company spend very little or no time trying to extract license fees from competitors, but just use the patent like some sort of legal scarecrow. Patents are supposed to be an incentive to innovate and this defensive use really is just a cost and does not serve to cause innovation.

I personally believe that software patents are primarily the tool of large companies with portfolios of patents which they cross-license with each other. Generally, it serve to keep competition out of the market and allows those with patents to push those without patents around or cut them out of markets entirely. A number of open source licenses are now dealing with software patent issues by creating incentives for participants not to litigate against each other. A focus on open standards is also another important way to try to keep innovation unencumbered by patents.

I am not against patents generally and I have worked in materials science and manufacturing technology companies where patents serve as a strong incentive for innovation and royalties provide a fair return for the investments. I just believe that the notion that software patents somehow help venture businesses is a red herring and that software patents are primarily a tool for software monopolies to stay keep the little guys out.

I am basing my opinion on personal experience. Your mileage may vary. I would be interested in the opinion of any VCs who feel strongly that software patents generally increase innovation and investment in venture businesses.


I have a small software start up and I went through exactly the kind of process you talk about here. I spent a great deal of time agonizing about whether to file a patent bearing in mind that it would be tremendously time consuming, expensive (bad when you have few resources) and distracting.

Eventually I decided to write my own patent, as much for the learning experience as anything else, but I wasted several months of time in which I could have been developing and more importantly selling software. To add more irony into the mix by the time I had finished the patent (at almost 200 pages) I had begun to design version 2 of my software, a design which superceded the core design of version 1 and the patent. Meaning essentially that my patent would be for something that I no longer used.

It became very clear that, despite the rhetoric that comes out of large corporations, creating and filing a patent was coming at the EXPENSE of innovation. Small businesses can generally do one or the other but both is difficult.

However, that being said, there are two large problems a small software company faces when trying to raise capital or make sales that I have both read about and encountered. That is, VCs and potential customers asking if you have patent "protection" (is that like mafia protection?) for your software and shying away from a relationship if you do not.

It seems that the question is almost "do you have any form of patent, weak or strong, with which to use as a shield or first line of defense to ward of potential predators?".

The problem for a small company is of course that the desperately dysfunctional patent system (surely the core of a patent is that it is for something that you have actually BUILT and can show!?) is desperately skewed in favor of deep pockets. With patent litigation currently estimated to be running at around $1 million - $2 million most small businesses (i.e. your, $1 million in capital, investments) cannot even begin to defend themselves, even if they have a "cast iron" patent as a shield if someone litigates.

Perhaps predator patent litigation is something that needs to be factored into software design going forward? Making systems fluid under the hood so that any component that happens to be litigated against (note i do not say infringing) can be easily swapped out and replaced?

Its a minefield for the start up entrepreneur (who need to read deeply about buzzwords like "VC", "patent", "start up" and understand before making decisions like this). What to do? You can help, use your influence whenever you can to push others with influence to pursue innovation over litigation. Development over claims writing. Engineer salaries over Lawyer fees (unless its my Lawyer who is a nice guy, in case hes reading ;).

What has happened in european parlament's not been the best piece of news posible. The project's just been rejected, but could be presented again. The best piece of news posible would have been the project being accepted with the ammendments that exclude computer programs from being patented.

Well, I use "best" from the point of view of somebody opossed to software patents, of course. :)

Joi, you should speak at Linux Bangalore 2005 in December this year.

First, Joi, I believe that most, if not all, of what you say is equally applicable to non-software patents and non-software startups.

But I agree that for the smallest startups, patents may not be the best use of their time or energy. It is difficult to say though, because these things can be very fact-specific. I am aware of someone who invented a new card game, and patented it immediately (only for the U.S.) because there is such a huge potential for others to steal the idea in that market, and it's impossible to make any headway whatsoever against entrenched interests without that clout. Although he would not have the money to litigate the patent, he would probably be able to extract some sort of settlement.

Filing only for your home country is much cheaper than filing all around the world. In general, it's like buying hotels in monopoly -- you can only get the protection that you can afford. Start small, with trade secrets or copyright, and when you have the means, you may desire the further protection that a patent can give you. The costs are mostly there because of the technical preparation, and the examination that's required in order to (at least try to) weed out old and obvious applications.

It would be great if patents were cheaper to get, and they may well get cheaper over time, as the processes involved are modernized. But there will always be a cost to analyzing and describing technical subject matter. As with all such things, there are different tools for parties of different financial means and different circumstances.

I see the biggest problem as the cost of litigation, and for the most part, it comes from our discovery process. I really think that this is where the most significant reforms should be made.

I would assert that software is slightly different. A lot of innovation in software is done by individuals, small companies or through collaborations in the public. (Open Source.) The innovations are generally incremental and patentable ideas come out through the process of working on the code which can be released over and over again as new versions. Revenue can be earned before there is significant investment and there is probably more money spent on marketing and distribution than on research. The process is very different.

Take drugs or synthetic materials as a counter example. Millions of dollars must be spent on research, equipment, trials, etc. before any revenue will be realized. The development period can take a VERY long time. Generally, the research will be done by larger institutions. There is also a number of good examples where one group does the research and another entity commercializes the product, rewarding the researchers through patent licenses.

In the case of software, I do not know of any significant organizations which do software R&D to produce patents to license, except a few edge cases.

Generally, software patents are used to litigate or as a defense, NOT as a normal way to earn licensing revenues from happy licensees. There are some examples of proprietary standards in consumer electronics that use patent pools. Some of these people might argue that it makes sense. Generally, I think these systems only work for rather large organizations. Again, there are edge cases.

You're right that the investment in software products nowadays usually doesn't have to be as great as for a drug. But it depends on the software. One of these days, we'll stop talking about the 'software industry' as if it actually had some meaning, and start talking about the operating systems industry, or the graphics engine industry, or the business applications industry, etc.

Some of these industries require almost as much R&D as more traditional industries. These would be the more technical areas, such as graphics, encryption, AI, etc.

But the patent system, IMHO, does not just incentivize innovation in the high-investment industries. It also incentivizes 3rd party investment in new ideas by newcomers who lack other capital. In such cases, the existence of a patent (or even a patentable idea) may serve as a 'signal' to the market of the value of an idea. But I digress...

It's true that there aren't that many companies yet that take full advantage of their licensing opportunities. But I believe that this is due to an outmoded understanding of the potential of intellectual property for licensing purposes. Traditionally, IP was viewed as a defensive regime; only recently have companies woken up to the fact that significant value can be gained by licensing their IP rather than attempting to implement it all themselves, especially in cases where they stumble on great ideas that are outside their core competency or market... So I see a much more vibrant market developing in the future for such licensing. This should create pressure towards further reform of the laws to reduce the transaction costs involved.

I think your drug example is a little flawed. Drugs need to be patented because the research/trials/etc is basically 'open-source' in that it has to be published in a reputable journal to be peer reviewed and accepted. As a result, there's no protection without the patent system. Look at the AIDS drugs discussions. It's never a question of whether a government CAN make a generic version of the drug but WHETHER they should.

If all software were open source, you'd need patents more than you do now. Reverse engineering of closed source software isn't that hard but it's still a lot tougher than opening a scientific journal or reading a patent application. You're right though that it can still be profitable to sell software without a monopoly protection.

The software patents discussion has me a bit confused in that the same people that say that 70+ year copyright is too long tend to say that 14+14 year copyright is too short (see Lessig's blog today). But a 20 year software patent is too much?

Anon: Yes. Patenting allows you to be more open about the research, I agree. I think the issue about generic drugs and IP protection for globally critical things like AIDS drugs is in a different league of discussion, although the innovation arguement is often used.

I read the post by Lessig. He didn't really say why he thought 14+14 was too radical...

So I think 80% of the problem with software patents is the system that is currently in place. It is expensive, can't handle the claims well and is abused. However, I'm not sure whether you could make a centralized patent office that manage something like this well. I would like to see a good proposal.

I am maybe too focused on my narrow field of experience, but I believe that some of the best software innovation is done in small groups with very little funding. I just can't see how you can justify spending a significant amount of your very limited resources on something as expensive and bureaucratic as patents when the whole team is basically doing CPR every day to keep the business alive...

In the medical industry (pharma, bio, and devices) we like patents. It isn't just that we are trying to protect our innovations; we are trying to manage our risk.

We spend three to twelve years developing a product with a life of five to fifteen years, spending tens or hundreds of millions of dollars. So what we get for the patent is a reduced risk of competition in the target market for a while - until the competition finds an alternate approach. This helps us justify the outrageous money we spend on the already very risky development process.

For products that are promising we will will issue a bunch of patents designed to close out similar designs/chemistry/genetics but there is ALWAYS an alternative way to achieve the same outcome. And our space WILL be invaded if it is profitable, and our competition will know since our clinical study results are published long before we get to sell the product.

So you buy patents to buy time. What happens is that you force your competition to do their own expensive development instead of riding on your coat-tails. The portfolio of patents will reduce the risk of someone competing with you in that first three to five years when you can charge premium. How long was Prozac the only SSRI, for instance?

A pretty simple biz analysis leads us to figure the value of the patent portfolio to protect a product as being primarily a protection for a period of about half of OUR developement timeline. We've already proved the concept workable, other people just have to figure out a different way of doing it, thus the assumption that it will take them half as long.

If we spend $100M to develop and $5M for patent protection, and ensure that the first five years belongs to us, it is a win. After five years there will be competitors, and the odds are they won't have the same complications, so we are unlikely to get much benefit. If, and only if, the product is still profitable are the patents maintained. We kill them off in quarterly triage when market share gets small unless they have strategic value or can be sold/licensed (there's a book worth of strategy to the license/sell/kill decision).

So in a portfolio of R&D projects we'll have a bunch of patent collections, growing as the projects look more promising, costing us about 5% steady state in each project. For us, it pays, because we are essentially buying guaranteed monopoly for a few years before someone else figures out how to achieve the same results with a different method and we end up in a competitive market. Remember that we have to pay for ALL of our development projects every year with the one or two products that are very profitable.

Why software companies think patents are useful is beyond me. You don't usually have twenty projects in the works, they won't take ten years to prove out, you don't have to do clinical studies for five more years to get permission to sell... etc.

I know why the established big players like patents, though. The more patents that are required, the more barriers to new entrants, and the more uncertainty there is for investors looking at new businesses that don't have IP... further decreasing competition. Plus, the more patents that go into the already underfunded patent system, the more bad patents get issued, increasing the amount of litigation which is a game only the established parties can win. So it's a good thing the software patents got shot down.

I'm just happy to see that much less new junk clogging the system.


thanks for a fantastic, and very succinct, description of the issues.

Your perspective as a VC is what's key, here. I did some lobbying against the pro-software-patent text for the EU CII Directive in Ireland, and I noticed that the VC angle was the other side's most effective FUD by far. Sadly, there are very few European investors (Laura Creighton being the notable exception) who comprehend how software patenting is not a viable strategy for small companies -- or at least are prepared to state this publicly.

Here's my short proposal:

a) No software patents, because it limit's the
free speech.

Think of general language as a programming language and
a speech recognition-abled robot as a computer and an
ordinary english text as source code. One does
need a license to distribute a patented "invention"
and web-publishing is a form of a distribution.
That's why software patents chill Open Source Software.

b) If the patent office "is competent enough" to do
prior art searches, then the patent office
is also competent enough to
provide a certification service. For
example, if a patent office has certified that
a product, let's call it NICEPRODUCT,
does not infringe any
patents or infringes only patents X Y and Z, then
that would be a FINAL DECISION. If it turns out that
some other patents are still infrnged, then the
manufacturer is not liable in any way, because,
it's the patent office's undone job. Besides,
they do a bad job even today, so, why not to
put some responsibility on them, right?

b) Monopoly based patents DO NOT WORK. The license should
be compulsory and the pricing should be like a tax:
always ONLY X%(X is a SINGLE CONSTANT and is determined
by law) of the FINAL SALES PRICE. So, if the patentable
invention is so valuable that the product can be
sold wiht high price and/or high quantities, then the
inventors/venture capitalists get their fair share. If
the licensed invention, which is a real kick-ass value
adder for some product, does not contribute so much to
the current product and the current product is not good
enoug(for whatever reason) for selling with high price
and/or big quantities, then the inventors/venture
capitalists also get their FAIR SHARE.

d) The predescribed X% of the final sales price get's distributed between ALL patent owners, regardless of their number. This way it doesn't chill incremental development(as the most usual way of doing development nowadays?) For example, let's
suppose that our sample product, NICEPRODUCT,
uses patent G, which is
derived from technologies with patents E and D.
Let's suppose that the technology, which is covered
by E, is derived from tehcnology that is covered by
L. In short:

OurProduct -> G
G->E and D

Patent fees in short:
owner of G gets 50% of X%
owner of D gets 25% of X%
owner of E gets 12.5% of X%
owner of L gets 12.5% of X%

You see:
less risks for the manufacturer, easyer adoption of
new(at least patented) technology, and, if combined with point C, there's also no way for anybody to come and do a "patent stickup". Let's not to mention the decreased mound of byrocracy and a possibility that real inventors get a good and real salary.

That's my proposal. And yes, NO SOFTWARE PATENTS, because it interferes with basic FREEDOM OF SPEECH.

OK, sorry for the numerous spelling mistakes, I'm
using Windows and I don't have the
Kate(an editor on Linux) spellchecker at hand.

7 TrackBacks

Listed below are links to blogs that reference this entry: One venture capitalist's view on software patents.

TrackBack URL for this entry:

I read this thread on Joi Ito's blog with interest. Both the post and the comments make some very interesting points about how the patent system actually functions in the marketplace. Here's an excerpt from Joi's post: I hear that... Read More

Fortunately we won't see in Europe early Read More

Contrasting views on “Intellectual Property” from Open Source Solutions for Small Business Problems
July 11, 2005 3:51 AM

Today, there’s a review in the New York Times about Hot Property, a new book by Pat Choate that essentially claims the US is in technology decline because of “piracy.” From the article: Choate says that while industrializing cou... Read More

Joi has a terrific post on why he, as a venture capitalist, is glad Europe rejected the software patent directive. Snippet: I personally believe that software patents are primarily the tool of large companies with portfolios of patents which they cross... Read More

Venture capitalist Joi Ito blogs his feelings on the European rejection of software patents. Ito has experience with companies in software, materials science and manufacturing technology and while he avers that he is a supporter of patents in general, ... Read More

I recently posted on Massachusetts's declaration to move to open document formats for government documents. Matt Asay raises several issues (innovation and customers, my possible confusion of patents and innovation, me being anti-Microsoft). One I've c... Read More