Monday, December 22, 2014

Big Data + Genomics ≠ Your Grandparent's Monoculture

Recently there was a really nice article in the NYT about how big data is being leveraged in the ag industry, Working the Land and the Data. There were some really great observations made in the article:

“we’ve got sensors on the combine, GPS data from satellites, cellular modems on self-driving tractors, apps for irrigation on iPhones...Tom Farms has genetically modified crops, cloud-computing systems and possibly soon drones, if Mr. Tom does not go with lasers on low-orbit satellites. All of these items will be sending their data for analysis on the cloud-computing systems that Tom Farms rented from Monsanto and other companies...Brent Schipper takes data readings from his combine every three seconds”

This certainly illustrates the volume, variety, and velocity of  big data I have previously discussed at my data analytics blog EconometricSense. And utilization of this data appears to have a payoff:

"better uses of data analysis have raised his return on investment to 21.2 percent, from 14 percent"

But there was some brief commentary contained in that article that I think could use some additional reflection:

"And there is another risk. There is an incentive to grow single crops to maximize the effectiveness of technology by growing them at the largest possible scale. Farmers with diverse crops and livestock would need many different systems. Smaller farmers without technology could also grow one crop, but they would not capture most of the gains."

This basically sounds like raising a concern about incentives for what some refer to as 'monoculture', or raising large acreages planted to single crops. Some are concerned about this because its like putting all of your eggs in one basket, for instance if there are environmental, pest, or pathological stresses that affect a single variety, then losses would be much greater than you might experience in a more diversified operation. There has always been a concern about market incentives reducing diversity in agricultural production, however, as I have mentioned before these concerns (about market incentives) may be somewhat exaggerated. In fact market forces can tend to be a driver of key aspects of diversity in agriculture.

In addition, one of the primary reasons that companies like John Deere and Monsanto are so interested in Big Data, is that there is money to be made in crop diversification. As I have mentioned before (See Agritalk Discussion of the impact of big data, biotechnology, and genomics on seed choice):

"the disruptions of new technology, big data and genomics (applications like FieldScripts, ACRES, MyJohnDeere or the new concept Kinze planters that switch hybrids on the go etc.) will require the market to continue to offer a range of choices in seeds and genetics to tailor to each producer's circumstances of time and place. There are numerous margins that growers look at when optimizing their seed choices and it will require a number of firms and seed choices to meet these needs as the industry's focus moves from the farm and field level to the data gathered by the row foot with each pass over the field."

(see also: What does the farmer say about seed choices)

"That's also why the market has driven companies to treat hybrid selection like a 'big data' problem and they are developing multivariate recommender systems as tools to assist in this (like ACRES and FieldScripts). The market's response to each individual producer's unique circumstances of time and place also ensures continued diversity of crop genetics planted."

The point is, as big data drives more diversity into every seed planted in every acre across every field, we may possibly begin to mitigate some of the risks and concerns traditionally associated with monoculture. So it is true, when you look across row after row and see only corn, you might technically call it 'monoculture' but its not your grandparent's monoculture.

See also:

Big Ag Meets Big Data (Part 1 & Part 2)
Big Data- Causality and Local Expertise are Key in Agronomic Applications
Big Ag and Big Data-Marc Bellemare
Other Big Data and Agricultural related Application Posts at EconometricSense

No comments: