Coffee science methodology Episode 3: Simplicity with Occam's razor

In this episode Occam's razor is introduced which is an important foundation for a good theory and examples of where this is not applied correctly is explained

Transcript

Occam's Razor

The first principle I would like to introduce is what is often referred to as Occam's Razor which urges you to always look for the most simple explanation rather than the most complicated if nothing more is gained by the complication of the explanation. It sounds so simple and true but this single principle is perhaps the most violated principle in the coffee community as it seems that many people are making a great effort to overcomplicate things rather than simplifying things, so it is definitely worth mentioning explicitly and call out as a separate principle to apply from our historical heritage of scientific methodology. 

William of Ockham is traditionally referred to with this concept but actually, already Aristotle came up with this principle:

“We may assume the superiority of the demonstration which derives from fewer postulates or hypotheses” 
Aristotle (384–322 BC) in Posterior Analytics

And William of Occam says:

“It is futile to do with more things that which can be done with fewer” 
William of Ockham (1287–1347 AC) in Summa Totius Logicae

Examples where this principle is violated in the coffee roasting community could be these

• 1) Most of my students arrive to my classes with a strong focus on Rate of Rise rather than just the overall time of different events of the roast. The RoR is not some hidden root cause for the roasting process that you need to look at independently to understand the roast because it is just a simple derivation of a much simpler aspect namely the shape of the bean curve that you already see even without calculating and plotting the Rate of Rise. You can use the RoR to predict how the roast will develop the next few minutes given the current speed or you can use it to classify different roast profiles on how fast they roast after 1st crack or other simple uses. But claims such as ‘I’m pushing the RoR mountain earlier in this roast’ is equivalent to saying ‘I push the time to 1st crack earlier in this roast’ and since referring to the simple time to 1st crack is much simpler than to RoR the time to 1st crack is preferred. 
• 2) Also, RoR can be reported in deg/15s, deg/30s or deg/min. Given that the roast logger software can optimize it’s own sampling rate independent on how the RoR is reported it is by far preferred by Occam to be reported in deg/min since this is the simplest expression and easier to use for for example future predictions of the roast. I have not seen any speedometers in cars using km/5 min or other strange time units not directly and intuitively useful by the driver.
• 3) Development time ratio in roasting education is another overcomplicating concept which we will deal with later because it also violates a lot of other aspects of a good theory (my central problem with this theory is I don’t know what it predicts at all). Most of the students I get on my online courses from around the world would tell me the DTR before the time for first crack and development time and most roast logger software would show you DTR rather than development time and you have to actively change it in the settings if you want to have development time displayed rather than DTR. No wonder newcomers to coffee roasting thinks DTR is more important than development time... Ockham would prefer to just report the roast in minutes for each phase and only when specifically need derived values can be calculated. For DTR I don’t know when it would ever be relevant where I see some situations where RoR is useful.
• 4) In a Youtube video with the title “A Rant: Espresso Ratios & Recipes” (https://youtu.be/45Ja8pJU73s) James Hoffman has a rant about brew ratio that I have also thought about and it was really relieving to see him annoyed with exactly this. I think this is another good example of an overcomplication of the concepts used in brewing theory. People would often tell you the ratio as if that is the central aspect of the recipe where I would like to just know how much coffee do you use and how much water goes through in order for me to directly make the coffee. The ratio I’m old enough to calculate myself IF I need it for a specific purpose. Just as Rate of Rise tells you something about the speed of the roast at different parts of the roast, the ratio tells you something about how concentrated a brew you make. There is a time and place for RoR and brew ratio but it should only be used if there is an explicit aspect of the roast and brew that you need to evaluate and never a substitution or default reference concepts. Ocham would prefer keeping things in minutes, degrees, grams of coffee and ml of water in most cases and only when specifically needed do these derived values for the purpose at hand. Which is probably needed much less than done by habit in the coffee community. There might be an element of ego sticking out for people using these concepts a lot. Doing these calculations gives you an aura as a scientist but people with background in science would see through this as an example of sophism as you take what sounds scientific just in order to persuade your audience to perceive you as a scientist where you are doing the opposite of a real scientist and thereby misleading your audience and put them in a basement of the cave with artificial light rather than out of the cave where the true light of knowledge is shining on the ideas behind the world itself to stay in Plato’s allegory.

5) Defaulting to calculating and using ratios and speeds in all situations rather than just using them in the very specific and only situations where they could be useful has another built-in problem: You lose information in both types of calculations! If you provide me with the recipe for a brew or show me a roast curve I could always calculate the ratio, Rate of Rise or Development Time Ratio myself but if you give me the ratio, Rate of Rise or Development Time ratio I can’t go back to the recipe of the brew or the roast course because I have lost information during the calculations of these derivatives. You have the same ratio in a 1 liter brew and a 200ml brew and you can have different RoR on roast going through the exact same process if the probes are different or placed differently. Likewise a roast with total roast time 10 min and first crack after 8 min gives you a DTR of 20% which is also the case for a roast with 20 min total roast time and first crack at 16 minutes. The fact that you lose information when calculating the speed of a process can be illustrated with mathematics by looking at the technicalities of finding the speed of a power function by applying the power rule which is beyond the scope to explain in this oral medium of a podcast but I have found good explanations on wikipedia and Kahn academy so please go to the show notes if you want to see how information is lost in the process of deriving the speed of a power function

Wikipedia: https://en.wikipedia.org/wiki/Power_rule

Kahn academy: https://www.khanacademy.org/math/old-ap-calculus-ab/ab-derivative-rules/ab-power-rule/e/power-rule-intro