The European Union (“EU”) recently concluded its investigation of the Bayer Monsanto transaction. As part of the remedy, Bayer has agreed to license to BASF its “entire global digital agriculture product portfolio and pipeline products to ensure continued competition on this emerging market.” According to the EU’s press release, “[d]igital agriculture uses public data such as satellite pictures and weather data as well as private data collected from farmers’ fields. It applies agronomic knowledge and algorithms to that data to recommend to farmers how to best manage their fields. For example, how many seeds to use, and on how much and when to use pesticide and fertiliser. This makes digital agriculture important, not only to farmers but also to the environment.”

The use of the word “license” is important; it is not, apparently, an assignment. The reason why it’s important is because Bayer can continue to use and market the intellectual property itself. At first blush, it would appear that the EU has created another “flavor” of the Bayer product that BASF can do with as it pleases.

At the outset, it is important to note that the intellectual property being licensed is not a patent. A patent describes clearly a particular art and grants the owner the ability to exclude others from practicing that art for a period of time. In exchange for that exclusivity, the world gets to see the art. A software license, or the code that executes an algorithm, is copyrightable. But a copyright protects the expression, not the idea. Moreover, software can be millions of lines long and will almost necessarily change over time. People’s needs change. Functionality can be added or deleted. Bugs are corrected. Software is a very dynamic product whereas the art in a patent is static. These are important differences.

Let’s say Microsoft is convicted of monopolization of the spread sheet market, and they are required to “license” Excel to someone who can sell it in competition with Microsoft’s product. At first, the two flavors of Excel sell equally well, perhaps even the divested version sells better because the purchaser can afford to sell the product for less as its sunk costs are less. Customers use the products and their needs change, adapt and grow. If I have just the right to sell the software and no knowledge of how it was built, it is very difficult for me to make any meaningful changes to the platform. Indeed, the more complicated the software is, the more likely I may destabilize it by altering it. In effect, with just a license, my “flavor” stagnates. It remains Windows 95 in a world where everyone is using Windows 10.

As the EU mentioned, there is a great deal of data associated with digital farming. It’s not only the weather, but the composition of a particular farmer’s soil, the atmosphere, the fertilizers, the timing of planting. It is also the type of seeds, the genetic composition of the seeds, how the seeds react in different environments. There is a great deal of data, and it’s not stagnant. To be useful, BASF would need access not only to historic data, but ongoing data. They would need access to the genetic code of the seeds, how that code interacts with the pesticide, how that pesticide is faring with the pests. And BASF would need “honest and complete” access to that data over time.

The digital farming product therefore is tied to the actual seeds, the traits and the pesticides; needs long term relationships with farmers; and the engineers who created the code in the first place. So long as the divestiture is of all of these things, freeing BASF from any tether to Bayer, the remedy is structural. If the license is just of the software, however, it would be an almost useless remedy because the government would have to impose a huge number of conditions on the licensing party to support the software in the hands of the competitive. And it would be inferior. The employees of Bayer would have little incentive to keep BASF up to date on the new and great functionality and analysis Bayer has discovered. Such a divestiture would become a behavioral remedy in practice. Software licenses (including big data) and patents are very different animals. You should never assume a software license (even an “assignment”) is necessarily a remedy as effective as a patent license.