It seems to me that this article, Don’t rush to machine learning, reminds us of an important rule that we often forget: start simple. Many times people try to overengineer a solution, when an 80%, just-good-enough solution will fit the purpose.

I’m in the middle of finalizing a plan for a short-term project that needs to be done quickly because it will only be useful for the next few months. And yet people continue to suggest additional features. As I keep saying: getting a minimally useful tool done quickly. We can always add on features later, but right now, having nothing isn’t helping you at all. Wouldn’t it be better to build a bicycle that at least helped you travel to your next destination, instead of trying to design a Cadillac that won’t be done until next year?