Reasoning from first principles is a powerful way to gain better understanding of complex issues and problems. By breaking things or events into more basic components, and tracing them to their logical origins, the so called first principles (i.e. the foundational propositions or assumptions), we can often reveal root causes and truly explain why things are the way they are. If the roots of a tree are the relatively small set of first principles, then this process is like tracing a particular leaf down its branches, then bigger trunks, and eventually arriving at the particular part of the roots that is responsible for the leaf.

But this explanatory power is mostly unidirectional: it does not work very well the other way. Trying to work the process in reverse, that is, starting from the set of first principles then deducing what will happen in a particular scenario, is essentially the work of predicting the future, which is nigh impossible. First principles come from the distillation of countless real world events and experiences, and are formulated abstractly via an inherently lossy process (in a way it’s similar to how a cryptographic hash function works). An infinite number of possibilities could stem from the same set of first principles. It’s like trying to predict exactly which leafs will grow from just the roots of the tree.

For example, it is reasonable to deduce the reasons that Google created Accelerated Mobile Pages to some set of first principles about how businesses always try to maximize profits, and that companies which rely primarily on ad-revenue from the mobile web will want to control how content is presented there. But to reason the opposite way, starting from those basic principles, and trying to predict the creation of AMP by Google, is equivalent to predicting what Google would do in general and not at all trivial.

So we should take care not to fall into this trap stemming from the asymmetrical power of how first principle thinking works. One of the ways that we are often misled is when we apply first principle thinking to hypothetical scenarios. It feels like we are going the “right way”, starting from something specific and trying to figure out its root causes within the abstract principles. But because the situation is hypothetical, we are actually not. This type of thinking is really just another form of starting from the roots of a tree and deducing how a future leaf will be, but in disguise. Reasoning this way only tells us that the hypothetical scenario can happen, not that it will happen.

This does not mean that we shouldn’t use first principle thinking in this direction, in a generative fashion. When done appropriately, this approach can offer great insights. Starting from first principles and going forward can give us valuable information on what scenarios are not likely to happen, i.e. the negative answers. This can often help eliminate large areas of future possibilities from our concern. And while it cannot reliably lead us to specific conclusions about the future, it can often show us rough areas of high likelihood events as well as trends about the future.

And even though reasoning from first principles to some specific hypothetical scenario does not guarantee that the future will happen that way, it does help paint a vision of the future that we can work towards. If some version of the future is desirable, and seems possible from first principles, then it is up to us to make it happen.