We're Doing External Innovation Wrong, But There's A Lesson From Evolution

Bill Joy of Sun Microsystems is credited with coining Joy’s Law of Management -- that no matter what company you are in, most smart people are outside your organization -- and it has stood for years as the maxim on the importance of partnering, licensing, and other forms of external innovation. There are plenty of positive anecdotes -- stories of companies that have struck it rich using external innovation. The problem is, there are also plenty of companies that have struck out. Remember Sun Microsystems? Despite their adherence to the maxim, they were outcompeted by other, more closed systems. So what does a data-driven review tell us -- is external innovation delivering the value for which it is touted?

First, let’s set some context by going back -- waaaay back -- to the formation of the earth (see figure 1). The great evolutionary biologist John Maynard-Smith was among the first to point out that you can think of evolution in a series of major transitions -- great leaps forward separated by hundreds of million or even billions of years. He also pointed out that each of these major transitions is an example of cooperation coupled with specialization to create a more effective whole. We have molecules working together to make simple cells, organelles working together to make complex cells, cell types working together to make organisms, jobs working together to make a society, and so on.

Now, viewed in this context, external innovation -- which is just cooperation across the boundaries of companies coupled with specialization -- must be really valuable… that’s the first conclusion. The second is that it may unfortunately take us a billion years to get it right.

Figure 1. Some major transitions in evolution

Let’s leave the earth forming 4.6 billion years ago and zoom in to the present day, and look at the largest ~30 pharmaceutical and biotech companies. I use pharmaceuticals and biotech because that’s the industry with which I’m most familiar, but the principles apply across industries. And for these companies, we’ll focus on R&D productivity -- the amount of innovation produced for a given level of investment, because external innovation is not a goal in and of itself, but rather, just one means to an end of getting the most bang for your buck out of R&D.

The R&D productivity of these ~30 companies is shown in figure 2, with each red dot representing a company. Perhaps not surprisingly, like many multiplicative processes, these dots are arrayed in a lognormal distribution, with a long upper tail. What is surprising, though, is how much variation there is between companies, with some doing hundreds of times better than others. To give you a sense of this, there’s something called the Gini index that measures the spread of a distribution. We can compare the spread of this distribution to that of another famous lognormal distribution -- the distribution of wealth in a country. A very flat country, like Sweden, might have a Gini index near 20. The distribution in figure 2 has a Gini index in the mid 40s -- somewhere between Turkmenistan and South Sudan, and actually right on par with Papua New Guinea. That gives you a sense that there’s a large disparity here -- some companies are doing really well with their R&D and others are not.

Figure 2. Distribution of R&D productivity for largest ~30 pharmaceutical and biotech companies

Now the question is, does usage of external innovation drive any of the enormous variation seen here? Are the companies at the top the ones that use external innovation the most?

The answer is no. As shown in figure 3, greater usage of external innovation -- across types -- is not correlated with better R&D productivity. The only exception may be large mergers. While conventional wisdom has viewed these as detrimental to R&D, a data-driven review actually found they reinvigorate R&D by bringing together the best science of both companies and at the same time eliminating low-value spend (see https://www.sciencedirect.com/science/article/pii/S1359644617301927).

Figure 3. No correlation between most types of external innovation and R&D productivity

So what’s going on here? Why don’t we see a general relationship between external innovation and R&D productivity when both Bill Joy and John Maynard-Smith say we should?

To answer that, first we have to understand what it is that really drives R&D productivity. We were able to show in a 2013 paper (https://www.nature.com/articles/nrd4164) that what really drives R&D productivity is just two things -- having access to great scientific information, and then actually listening to that information when making decisions.

There’s no question pursuing more external innovation opens the aperture for access to scientific information. Figure 4 shows Joy’s Law in action. It is a map of all the scientific publications in just one disease area (in this case, rheumatoid arthritis) with each dot a publication and each line a shared co-authorship. The footprint of one large pharmaceutical company is shown in green. It’s a pretty big single contribution, but you’d be nuts to think all the good science is within that little green dot.

Figure 4. Map of publications in rheumatoid arthritis

But unfortunately there’s a flip side to Joy’s Law, which I’ll call Joy’s corollary: most bad ideas are also outside your organization, and it’s very hard sometimes to know which is which. So to distinguish between the good and the bad, companies rely on internal scientists to make sense of all those external ideas. And this brings us to the crux of why there’s a breakdown between external innovation and R&D productivity: even if you have access to great scientific information, you still have to listen to that information when you make decisions. This is where most companies stumble.

Usually internal scientists can become absolute heroes by creating internal innovations, but the personal rewards for spending time evaluating external ideas are not nearly commensurate. This means time spent evaluating external ideas, except insofar as it helps develop internal ideas, pulls time from what can really help these scientists personally get ahead. Even worse, sometimes these external ideas are perceived not just a distraction, but as competition. There are stories from some companies of internal scientists being reluctant to bring in external projects when it meant they would have to kill their own similar (but inferior) internal project. In many cases, scientists still do the right thing despite these issues, but it’s a bad idea to set up company incentives in a way that creates these issues, actually making it harder for scientists to do the right thing.

In a nutshell, John Maynard-Smith captured all those years ago what’s going on in external innovation today. External innovation should work: bringing together complementary elements can definitely make 1 + 1 = 3. But to get it to actually work, we have to think about it not at the company level, but at the level of the individual scientists who are doing the cooperation with the outside world, and design the interaction in such a way that they personally do just as well or better by cooperating -- becoming true heroes for cultivating external ideas.

For more on this, please see my TED talk at: https://www.ted.com/talks/michael_ringel_external_innovation_basics_from_an_r_d_expert