I spent a lot of time over my holiday break thinking about the pharmaceutical discovery and development process. Specifically, I was preoccupied with the reasons for why it takes so long and costs so much to get a drug to market.

There’s a lot of passion. Smart people disagree on the real cost of discovering a drug, but everyone tends to agree that it’s insanely high. I actually believe the $1B number bandied about isn’t far off, and I also believe that the cost of failure is a fair part of the number of the costs of successes. The time for successes isn’t hard to argue though…seventeen years.

The success of “open source” licensing in changing software and culture gets a lot of folks excited about open source biology. Now, I believe in open science. And I have made my career choice in open science. I’m a believer in the power of open approaches.

But I am very skeptical that the law is the key, single solution here. Mainly because I believe the law is only a symptom of the big problem. I do believe that the commons itself is the solution, but not because of intellectual property.

It’s the solution because it’s an incredibly efficient way to generate knowledge rapidly, at low costs. And knowledge is what we need. A commons is the infrastructure for distributed collaboration and innovation in the life sciences, and we should be thinking about the law in terms of how to expose and integrate as much knowledge as possible at the lowest possible transaction cost.

The primary reason drugs are so expensive to discover and develop is that we just don’t know very much about the human body. The problem isn’t the law. To paraphrase Bill Gates, the problem is complexity (read the whole thing, it’s worth it).

This has got me thinking about the knowledge gap – the things we don’t know that cause us to fail so regularly at discovering drugs – and the way that gap interacts with the commons of knowledge and research tools.

More after the jump.

Complexity is the core problem. Human bodies are so mindbogglingly complex that we can’t accurately predict…well, very much of anything about them. There is a knowledge gap a mile wide and ten miles deep between where we are today and a world in which it costs less in terms of time and money to get a drug to market.

We don’t know how to accurately deliver a drug into a cell. We don’t know how to translate knowledge about a gene into an understanding of disease (for example, the Huntingtin gene was characterized in the 1980s – and we still don’t know how the disease works, much less how to intervene). We don’t know very much at all about how toxicity works in people. That’s a major knowledge gap.

As a result, the best way we have to test drugs is to give them to people and see what happens.

On top of that, as a society, we have little tolerance for managing this knowledge gap. Lots of drugs we take for granted today would never succeed in today’s regulatory environment. Aspirin, for example, would likely fail due to stomach bleeding problems (it would be compared to tylenol, which does not have similar numbers, and found lacking). This is a child of the knowledge gap. We can’t look at a drug and understand its impact…

The solution to the knowledge gap has, to date, been to simply take more shots on goal. We throw eight million potential drug compounds at a target and check to see what sticks, then we cull the field, again and again, tweaking here and there.

It’s like playing roulette and winning by betting on every square, then patenting the one that wins and extracting high rents from it. Patents are much less the problem in such a scenario than the fact that we are playing roulette. Changing the odds is the better answer – it lowers the pressure to rely on patents!

It’s the abject complexity of the human system and the reality of the knowledge gap about the system. Human bodies make microprocessors look like children’s toys in terms of complexity. And those bodies exist in a constantly changing set of environmental factors.

Complexity is a problem in another place, which is the structure of the modern large pharmaceutical company. Byzantine is a nice word for those corporate structures. At every step this game takes expensive, highly trained researchers, reagents, machines, and the overall infrastructure of a massive pharmaceutical company. On top of that you overlay regular M&A activity, regular organizational overhauls (this week, we’re organized into therapeutic areas, but next week, we’ll be organized by function, and the week after that we’ll be creating virtual centers of excellence!).

Complexity is the problem both in terms of our understanding of bodies and drugs and in terms of reworking the models around discovery.This system regularly and utterly defeats the best efforts of many entrepeneurs and policy reformers to change things for the better. There’s a beach full of startups poised to change the industry through technology, and the NIH regularly posts roadmaps, blueprints, and other five-year plans that seem to generate lots more data than information. No one involved is evil, but there is so much inertia and expense and above all bureaucracy involved that it’s just damn hard to make a change.

One of the reasons I believe so deeply in the commons approach (by which i mean: contractually constructed regimes that tilt the field towards sharing and reuse, technological enablements that make public knowledge easy to find and use, and default policy rules that create incentives to share and reuse) is that I think it is one of the only non-miraculous ways to defeat complexity. If we can get more people working on individual issues – which are each alone not so complex – and the outputs of research snap together, and smart people can work on the compiled output as well – then it stands to reason that the odds of meaningful discoveries increase in spite of overall systemic complexity.

This is not easy as far as solutions go. It requires open access to content, journals and databases both. It requires that database creators think about their products as existing in a network, and provide hooks for the network, not just query access. It requires that funders pay for biobanks to store research tools. It requires that pharmaceutical companies take a hard look at their private assets and build some trust in entities that make sharing possible. It requires that scientists share their stuff (this is the elephant in the lab, frankly). It requires that universities track sharing as a metric of scientific and societal impact.

It is not easy. But it is, in a way, a very simple change. It just requires the flipping of a switch, from a default rule of “sharing doesn’t matter” to one of “sharing matters enormously”. It is as easy, and as hard, as the NIH mandate on open access. It’s a matter of willpower.

If we’re going to attack the cost of drug creation and marketing, we have to attack the failures at the source – the knowledge gap created by complexity. Creating a robust public domain and knowledge commons – with the attendant increase in scientists who have the freedom and tools to practice collaborative science, all over the world -is one of the only clear methods we have at our disposal.

And if we can actually get the price point down to $100M, or $50M, the game is changed forever. VCs can fund a drug, as can foundations, at that price point. Prize models suddenly become very, very workable. And big pharma finally would see meaningful competition.

Complexity is the enemy. Distributed innovation, built on a commons, is a strong tonic against that enemy.