The beauty and curse of farming in the technological age comes in the form of 0’s and 1’s. There never seems to be an end to the measurables we collect and breakdown on a farm.
As our capabilities grow in terms of systems and software, this mound of data continues to grow.
But how do we know what is important?
Many seasoned herdsmen that I work with often bemoan just how hard it is to train the next generation of farm workers. “They just don’t have cow sense.” “They’re too deep in the numbers, and not the cows.” “Why can’t they just see it?”
What they are describing is a little-known skill they already possess known as Signal to Noise Positivity (SNP). Our seasoned herdsmen have developed an unconscious skill that allows them to differentiate between meaningful information (signal) and irrelevant information (noise). However, they had a significant advantage of developing this skillset during a time when the “noise” or extra irrelevant data was much less overwhelming.
Our farms are full of positive noise and signals that can be valuable in animal management. Feed intakes, milk production, lbs. of solids, rumination, SCC, milk deviation, etc. etc. all represent measures that indicate if things are headed in the right or wrong direction.
However, each of these factors is not important in each situation. Irrelevant noise from time to time causes new decision makers to make incorrect decisions based upon that noise rather than the more meaningful signals.
Taken together Signal and Noise Ratio (SNR) indicates whether positive noise is more or less likely to stick out. A higher SNR, or more meaningful noise vs irrelevant noise, means the decision maker is more likely to ID the important data whereas a low SNR means more confusion in the data.
But on today’s operations it is extremely difficult to learn how to increase a SNR and use it as our data pool, and subsequent noise grows.
For example, your herdsmen today may work a fresh pen and ID a sick cow with the following information; 7 DIM, rapid breathing, temp is 103 on the parlor meter, milk production is down 30 lbs., her ears are droopy, her rumination is down 40%, her eating time is 2%, and her activity is 50% what it was yesterday. These symptoms when taken together could be indicative of 5+ infectious diseases in a fresh cow and can rapidly confuse new workers.
However, an experienced herdsmen knows the most important part of all this noise is the 7 DIM signal which limits the likely diagnosis to only 2 or 3 possibilities. Their unique SNP ability filters the signal from the noise so no matter how much data we pile on top of the signal they can ignore the unimportant noise.
So how can we improve the SNP for new workers?
First, we need to eliminate multitasking. This only adds noise to the scenario and scatters attention. When working sick cows or doing other health tasks, new employees need to focus upon that singular task until deemed to have developed a sufficient SNP.
Second, we need to encourage monitoring of longer-term trends and react less to the hour to hour or day to day fluctuations. Certainly, there are cases that will change hour to hour but when training we need to help employees understand common cases and that cows generally will trend toward “healthy” or “sick” no matter how complex or simple our data collection system is.
Third, we need to expose these individuals to training from others who have advanced SNP skills. Many times, this involves an outside consultant such as your herd veterinarian who can use hands on training and also create SOP programs to help the individual navigate the noise.
Signals of disease are not always strong. However, by reducing noise for new team members we can increase the correct disease diagnosis while still implementing the latest in smart technology.
