There are two major approaches to making an argument: reasoning by analogy and reasoning by first principles. In reasoning by analogy, you start with a premise that has been accepted as true, then show how another premise (the conclusion) must also be true. In reasoning by first principles, we begin with basic axioms that no one would dispute (what we call first principles), then reason from there to our final conclusion.
These are two very different ways of approaching complex problems. Sometimes, you can use them in combination to find answers to challenging questions. But, starting an argument with first principles is the greatly superior choice.
Reasoning By First Principles VS Reasoning by Analogy
When formulating an argument, or trying to solve a problem, the way in which you think about the problem can make all the difference in the effectiveness of your solution. In this article, we’ll take a look at two strategies – first principles reasoning and analogy – and see how they are used in arguments, as well as how and when they work best. By understanding these strategies and their differences, you’ll be able to make better arguments and more effectively tackle complex problems like never before!
While analyzing a problem, to develop an effective solution, your approach will differ according to whether the problem space is familiar or not. First principles reasoning involves analyzing the problem from first principles (i.e., from the most fundamental assumptions), whereas analogy-based reasoning takes into account things such as analogies and similarities between the current situation and past situations that have already been analyzed successfully, as well as existing solutions that may be applicable in your situation.
If you want to reach a conclusion but lack sufficient information to do so, you must use reductio ad absurdum (first principles) reasoning to get there by working backward from your desired conclusion through successive steps of reasons why until you reach a self-evident truth that makes it possible for you to deduce your desired conclusion. To continue with our syllogism, we would start with reasoning by analogy and reason backwards from our analogy until we reach first principles logic (if p is A and A is B, then p is B). Once we reach first principles logic in our hypothetical example above, it should be clear that if All P are B then Some P are B; therefore we have successfully inferred Some Q are B.
Reasoning by Analogies Alone Can Lead You Down the Wrong Path
Analogies are often hard to counter, and can be convincing. Consider that A is to B, as C is to D; therefore, if B is a good solution for A, then D may just work for C, too. Of course, analogies are limited in their scope because they’re ultra dependent on similarity. Once you get past superficial comparisons, such as bicycles and cars both having wheels, analogies lose their force.
It’s easy to get locked into thinking that something must be a certain way just because it resembles something else. That’s why analogies can lead you down paths you don’t want to go. Analogies can prevent you from thinking clearly about new information and challenging your own assumptions. You aren’t going to learn anything useful by simply staying in your comfort zone.
Let’s say you’re trying to decide whether or not to invest in an idea for a startup. You’re considering whether it will succeed based on how similar ideas have performed in the past, using other startups as analogies. While it may seem like a sound reasoning strategy at first, analogies can often do nothing but cloud your judgment. Simply reasoning by analogies can mislead you into thinking that something is true simply because it resembles another situation where things were true. But, in many cases, there may be important differences between those two business situations which would make them behave quite differently in reality!
In our algorithm-happy world in which we try to lean on massive piles of data to cope with making big decisions, a lot of these algorithms often assume analogies. As powerful as machine learning can be, oftentimes what these advanced AI program spit out is going to be misleading. After all, these programs are only as good as the data they are given, and if that data is being analyzed under flawed premises, it’s actually useless.
Machine Learning is Only as Good as the Programming Involved
So, don’t blame machine learning for the strange ad choices you’re being given when you’re web surfing. You can’t fault AI for suggesting songs out of left field on your favorite streaming platform. Most algorithms are based on analogies crafted from making millions to billions of comparative analyses in the space of mere seconds. Now, for things as relatively inconsequential as suggesting the next TikTok video, or song for your playlist, analogies can be perfectly adequate.
But, if you’re looking for machine learning to make investments or solid arguments for making specific life-altering decisions, you’re to often relying on reasoning by analogy, albeit analogies given absurd mountains of data to refine. It’s not the computer’s fault if it spits out a ridiculous result; it’s whoever programmed it. Considering that AI could be replacing court judges sooner than later, and deciding if you get a home mortgage or an apartment lease, or any other number of life-altering decisions, keep this in mind when you get an unexpected verdict.
So, if you’re programming one of these machine learning algorithms, be sure not to start with analogies; start with first principles and use analogies to help streamline the decision making process. Even if you’re not a programmer, which is most likely if you’re reading this essay, you should follow the same line of reasoning. Start with what you know is true and build your argument out on facts, only using analogies only as guide markers, not absolute apples-to-apples comparisons.
While using analogies isn’t always wrong, as they can often help clarify arguments, they shouldn’t be the basis of your argument. Analogies should only be used after you’ve established your position with first principles or unarguable facts. Unfortunately, in today’s world being taken over by machine learning algorithms, it’s going to be harder than ever to think this way, as many decisions that will be made in coming decades likely will be made thanks to flawed programming and an over-reliance on apples-to-oranges data points.