Geeks and Tweaks: What Computer Programming Contests Can Teach Us About Innovation

Kal Raustiala, a professor at UCLA Law School and the UCLA International Institute, and?Chris Sprigman, a professor at the University of Virginia Law School, are?experts in?counterfeiting and intellectual property. They have been?guest-blogging for us about copyright issues. Last time, they wrote about the roles of “tweakers” and “pioneers” in the innovation world; today, they expand on the topic.

Geeks and Tweaks: What Computer Programming Contests Can Teach Us About Innovation
By Kal Raustiala and Chris Sprigman

We talked in our last post about two different kinds of innovators: Pioneers and Tweakers. Pioneers are the Thomas Edisons of the world – the people who fit our romantic image of the lonely genius – whereas Tweakers are the ones who transform Big Ideas into brilliant products by reworking and refining them.

Intellectual property law decidedly favors Pioneers over Tweakers. We think this is a questionable strategy. At the very least, we ought to think about what the right balance is between the two broad forms of innovation, and how we can design the right incentives to promote that balance.

An interesting window into this question is a longtime fixture of geek culture – the computer programming contest.? MathWorks, a Natick, Massachusetts firm, has sponsored a series of online programming contests to promote their MATLAB programming language. These typically require contestants to write a program that solves a single difficult math problem in the least amount of time.

One example would be the classic traveling salesman problem, which asks the contestants to find the shortest possible round trip a salesman can make through a given list of cities, visiting each city only once. Contestants in this version of the contest must write computer code to calculate the shortest trip, and then submit that code to the contest website. Contestants can submit as many entries as they like over a period of several days.? Each is immediately scored, and the rankings made continually visible to all.

At the end of the contest, the winner receives … a MathWorks t-shirt and public acknowledgment of his or her victory.? That’s it.? And for this, quite a few highly-skilled people will spend a lot of time – sometimes, more than a hundred hours – writing code.

But here’s an even more surprising twist: after a short initial period of “darkness,” where the submitted code is hidden, the contest is played out in “daylight” – i.e., all of the contestants get to see each others’ code.? And they not only get to see – they are allowed, indeed encouraged, to take. These rules lead to a innovation environment similar in some ways to what we described in our last post about? football.? Some contestants are Pioneers – they work out a fundamental insight that helps address the problem, and submit code embodying it.? Others are Tweakers: they take code from Pioneers, improve it and resubmit it.

As more and more Tweakers wring the flaws out of a Pioneer’s code, the solutions to the problem get better and better.? And more subtly, as the Tweakers push any particular Pioneer’s solution toward its best implementation, the limitations of the Pioneer’s original insight become apparent.? In this way, the Tweakers help to prepare the ground for the next Pioneer – someone who comes in with radically different code that avoids the bottleneck that limited the performance of the previous best solution.

OK, you say, so the Tweakers create some value.? But doesn’t any set of rules that lets the Tweakers loose crush the incentive to be a Pioneer?? Why would anyone want to work out a pioneering approach to a math contest problem if a Tweaker can simply take it, fiddle with it a bit and leap ahead in the contest?

Ned Gulley, the MathWorks guru in charge of the contests, has suggested an answer:

We find that tweaking is the thing our contestants most often complain about, and at the same time it is the feature that keeps them coming back for more. Our discussion boards swirl with questions like this:

  • Who deserves the most credit for this code?
  • Who is a big contributor and who is “just a tweaker”?
  • What is the difference between a significant change and a tweak?

These kinds of questions bedevil real-world software projects. There seems to be a cultural predisposition to find and glorify the (often mythical) breakthroughs of a lone genius. Since this model doesn’t always match reality, these questions don’t have satisfying answers. Happily, the contest framework acts as a solvent that minimizes this kind of I-did-more-than-you-did bickering and maximizes fruitful collaboration among many parties.

We find several noteworthy points in Gulley’s comments.? At the most general level, there is nothing inherently unjust about tweaking.? The MathWorks games show that although participants may sometimes complain about tweaking, they by and large accept it when they know in advance that it’s part of the rules.

Football is not that different.? Rich Rodriguez of Michigan runs summer camps where coaches can come to learn his implementation of the spread-option.? And a recent NY Times Magazine article describes how the New York Jets head coach Rex Ryan‘s training camps have become a Mecca for coaches seeking inspiration:

This is all why, throughout this off-season, springing up like gladiolus along the sidelines of Florham Park[, New Jersey - the Jets' practice facility], were dozens of coaches in polo shirts and twill slacks, with return airline tickets to Indiana or Hawaii on their hotel bureaus. One week, Jon Gruden, the broadcaster and former Raiders coach, came up from Florida to take the Ryan cure. Then it was Nick Saban of the University of Alabama, college football’s defending national champion, reviewing blitzes. “We’re all copycats,” Saban says. “I haven’t invented anything in this business. I’ve always watched what Rex does.”

In football, coaches are free to copy.? As a consequence, we see hardly any moral handwringing about copying because it’s part of the background culture and people expect it to happen.? In this way, we see that sometimes the rules don’t follow morality, but rather that morality follows the rules – i.e., what’s normal is moral.

The same is true of? the MathWorks contests.? Not all contestants like Tweakers, but they accept them.? As participants gain experience with the contests, their views of tweaking appear to shift.? They begin to see tweaking not simply as copying, but as innovation.

So how can legal rules encourage a healthy balance between pioneering and tweaking? We wrote previously about how patent law only weakly promotes tweaking. Copyright’s treatment of Tweakers, however, is far worse. Copyright owners not only have the exclusive right to copy, distribute and perform their works.? They also have the exclusive right to make what the copyright law calls “derivative works” – i.e., tweaks.

So copyright has a clear policy: No Tweakers Allowed.? If you write a novel, I can’t easily tweak it by transporting the characters into a different time or place.? Indeed, in 2001, Alice Randall famously tried just that with The Wind Done Gone, a tweaking of Margaret Mitchell‘s classic Gone With the Wind, but now written from a slave’s point of view. Mitchell’s estate sued Randall; a lower court issued an injunction against publication, later lifted by an appellate court. The case was only dropped after a large donation was made to a historically African-American college by Randall’s publisher, which now had a best-seller on its hands and could easily afford to do so.

Of course, nothing stops a Tweaker from getting permission in advance from the Pioneer.? Unfortunately, however, copyright law has made that increasingly difficult.? The copyright law in general provides very little information that Tweakers can use to find copyright owners.? There is no database of copyrights that a Tweaker can search, in order to strike a bargain with the right Pioneer.

The law didn’t always make it so difficult to negotiate, however.? From the first copyright law in 1790 until 1978, authors were required to register with the US Copyright Office, or, later, provide notice of copyright on all published copies of their works.? And rightsholders were required to re-register after a relatively short initial term of protection.? The result was a public record – not perfect, but very helpful – of who owned what.? And this made it easier for Tweakers to seek permission.

Those requirements are gone now, and along with them the copyright law jettisoned its best encouragement to Tweakers.? That should change.? We should at the very least consider re-imposing rules designed to help Tweakers find copyright owners.? But more fundamentally, we should consider changing copyright to make it more like patent.? If you improve my novel – say, by reworking it into an appealing screenplay – then you should be able to publish what you’ve done, so long as your changes are substantial.? But I should also get paid for what I’ve contributed.? In short, we should have a system where the Tweaker’s work also benefits the Pioneer.

TAGS:

Leave A Comment

Comments are moderated and generally will be posted if they are on-topic and not abusive.

 

COMMENTS: 23

View All Comments »
  1. Drill-Baby-Drill Drill Team says:

    The Real Pioneer was the inventor of the original Computer Program coding language.
    Everyone else just plays by his rules, rearranges his code and speaks his language. They are Tweakers.

    Stephen Jobs is a Pioneer. The Army of App Writers are Tweakers.

    Lewis and Clark are Pioneers. Oregon Trail immigrants are Tweakers.

    There are 99 Tweakers for One Pioneer. Genius comes once in a generation…if that. Most of us should stay with the flock for safety.

    Thumb up 0 Thumb down 1
  2. Mike B says:

    Not to defend the current copyright system, but there is nothing that bars one from completely appropriating the functional elements of a copyrighted work and re-creating a technically original work that borrows heavily from the original.

    There are several complications that make tweaking less feasible in many real world copyright situations. Where the costs of creating an original work is high, tweaking might become greatly preferred to creating said original work. Yes both the pioneers and tweakers would profit, but there might be less overall innovation. Consider a video game that costs 10′s of millions of dollars to develop. Tweakers could release mods or expansions and both parties could profit, but where one has the legal option to tweak high costs of pioneering would make endless tweaking with few pioneers the natural outcome.

    This is already seen in many areas of scientific discovery where almost all high risk basic research is carried out by academia of the Government. Even in a patent protected regime many drug companies prefer to tweak their existing drugs rather than try to develop new ones. tweaking of all kinds is simply lower risk and lower cost.

    If one takes the ability to tweak off the table or restricts it, then one will naturally get more pioneers. Yes both tweakers and pioneers are necessary, but there is a healthy mix of the two that will achieve optimum innovation. Making tweaking too attractive could upset the balance and make us worse off than the current setup.

    Also the changes to Copyright Law in the 1970′s were a result of European ideals regarding the sacred rights of the author. While they have worked to the advantage of Big Media, the intent was to allow authors to have complete control of their works for moral, not profit, reasons. The idea is that unauthorized derivative works can insult , devalue or defame the author’s original vision. Think of those popular bumper stickers of Calvin (of Hobbs fame) urinating on a automaker logo. The author objected to all forms of merchandising so such unauthorized works effectivly damaged his brand.

    Authors are not always profit motivated individuals and can go nuts when their work/”babies” are used in ways that they did not intend. Our current Copyright system now bends over backwards to accommodate such feelings.

    Thumb up 0 Thumb down 0
  3. Andrew says:

    Wonder what innovation experts like Clayton Christensen and Jeff Dyer would say about this.

    Thumb up 0 Thumb down 0
  4. Alan says:

    The analysis is good, the idea to improve ways of locating copyright holders also good — but to make copyright law more like patent law? Bad idea. I don’t want someone suing me over, say, this comment I’m now posting, based on allegations that it’s a tweaked version of what they wrote somewhere else. I shudder to imagine the copyright counterpoint of today’s patent trolls — companies that exist purely to buy up patents and then file infringement suits against companies that actually produce useful work.

    Current copyright law has actually been used in innovative ways to reward Tweakers: open source. Copyright holders (the pioneers) license their works in ways that permit derivative works (tweaking) to be made by anyone so long as those derivative works are licensed and made available under the same terms as the original code. The reward for the Pioneers is the resulting flood of improvements they didn’t have to spend their own resources on, and the rewards for the Tweakers are, as you mentioned, the ability to bring a product to the level they need it to be at without having to envision it and build it out from scratch. And they don’t need to pay royalties up the chain of provenance.

    Of course, the open source model is better suited to utilitarian works like software or encyclopedias, than to purely creative works like novels or music — though there have been efforts to make it work in the latter realm too.

    Thumb up 0 Thumb down 0
  5. Ian Kemmish says:

    I suspect that the rules of this particular contest attract (through accident or design) the two categories of programmer described, and no others. It’s hard to imagine anyone who used to go in for the Obfuscated C Contest being attracted by this, for example.

    Or me. I took, and gave my colleagues, large chunks to do, and made sure that each was wholly responsible for his or her code. When a panicky US customer calls up with an urgent problem at clocking off time in the UK, the last thing you want is to start assembling a team, none of whom completely understands the code, or worse yet, start debugging Open Source contributions.

    As far as I’m concerned, if you’re top in the benchmarks and the customers are happy, then your code is fast enough and no further tweaks are necessary (and indeed they will only introduce flakiness). If the customers aren’t happy, then tweaking is needed, but the person best placed to do it is the person who is responsible for the code (well, maybe with a few hints).

    I’ve just remembered that for the first four years of my career, I worked in the APL community, which has a very similar approach to that described in the article. I pioneered code for others, and I tweaked others’ code. It was fun. But it wasn’t industry. It was the other approach which made my code rock solid and fast, me wealthy and my customers happy.

    Although I’d love to hear Fred Brooks’ opinion….

    Thumb up 0 Thumb down 0
  6. Avinash Dixit says:

    This reminds me of Thomas Kuhn’s famous and often misunderstood account of progress in science. Most of the time, research activity is tweaking: incremental progress within the established tradition. Kuhn calls this normal science. That occasionally brings to light anomalies that prepare the stage for a revolution, or a change of paradigm. Most people have misunderstood Kuhn’s message; they undervalue normal science and glorify paradigm shifts. But Kuhn was very clear about the importance of normal science in the process. Read his brilliant essay “The Essential Tension” in a book with the same title.

    Thumb up 0 Thumb down 0
  7. Alex says:

    I feel the issues brought up by these analysis are heartily addressed by the Creative Commons approach of sharing (http://www.creativecommons.org). I would much appreciate an analysis of this system.

    Thumb up 0 Thumb down 0
  8. Eric M. Jones says:

    @- Avinash Dixit,

    Right on.

    I am puzzled why anyone thinks these two categories are important or even generally exist. True, if they exist, many are more one than the other, although this may change with time and circumstance and definition.

    Shifting from one technological area to another tends to move one from a tweaker to a geek. Many tweakers consult as geeks…Geeks usually don’t know a different field well enough to consult as tweakers.

    Thumb up 0 Thumb down 0