Great Analytic Organizations Often Employ A Healthy Dose of Bullshit
You Can Learn How, ASSUMING You Keep The Right Perspective
I am not a big fan of cliche’ or bullshit. Often the two are synonymous. More often though — people take them too far. Perhaps this article will as well, but hopefully it will help you find the value of bullshit.
For our argument, bullshit is synonymous with made up numbers. Many organizations struggle with numbers, estimates, and assumptions that are essentially made-up. And in the common vernacular they qualify as bullshit.
Some Organizations Need This Sign, Others Need To Let It Go
Analysts and data-driven professionals are often inclined to fight all bullshit and justifiably so. Organizations should never use made-up assumptions when real data is available. Failure to use real data comes down to ignorance, bad intentions, or impatience — none are acceptable.
Some organizations have a phobia on assuming. They bought into that popular refrain — assuming makes an ass out of you and me. To them, data-driven assumptions aren’t considered assumptions at all. They are just data. So let’s cut the bullshit…
When you use historic data to predict the future, whether or not you want to admit it, you are making an assumption.
While certain academic members of the analytic set will allow for this, they draw the line when an assumption is based on no data. If someone is making a choice to ignore data, I have no issue with that line. However, if truly no data or no meaningful data exists, the assumption still needs to be made. It most likely will, but in those organizations suffering from overzealous purists, it will likely not be made public. This is a both a problem and scientific hypocrisy.
Artificial Intelligence, Machine Learning, and Adaptive Algorithms Are Often Powered By Bullshit
The height of analytic science right now is artificial intelligence. It is a field that often engages in the “training” of models and algorithms. At the onset of this training, the model is given starting weights (aka assumptions). While these assumptions may be educated by real data, quite often they are not. Real data can speed the number of cycles needed to develop an optimized function, but our AI Overlords can start with a random number. Coming back to our purist data-nazis, this would mean they start with bullshit. Ah, the hypocrisy.
Our Example Provides The Hook — Learning
Our AI model, your organization, and anyone not engaged in meaningless bullshit is learning! Learning starts with assumptions. And if there is truly no data (or little meaningful data) available, you still need an assumption to start the learning process. It is not bullshit… although maybe it is fertilizer? Or perhaps I am getting too wrapped up in cliche’? Or perhaps we need to better distinguish our shit…?
Horses Have Manure Much Higher In Nitrogen Than Cows
Bullshit was a term I came to later in life. My dad favored horseshit. Many people, but plenty of farmers would agree.
In other words, horseshit is better fertilizer than bullshit. So perhaps we can make that distinctions? Others already have. If we accept their description, that bullshit is known to be false but horseshit is only later determined to be so (my summary), perhaps we should adopt the new nomenclature as well?
So cut the bullshit, but don’t be chickenshit to try some horseshit. It is the only way you and your company will learn… shit.
Perhaps that was too far? But I think you get the point or at least you have a new perspective. Consider it a little fertilizer… the equine kind.
Analytic organizations boldly make assumptions with the best available data. But when no data exists, the greatest organizations recognize that those assumptions still must (and will) be made. While they may be horseshit, those assumptions are just the starting conditions for great organization to begin their learning process. They drive real growth but only when transparent to the entire organization.
Thanks for reading! Now go get shit done!
