## Good Taste is the Combination of Both a Long and a Sensitive Feedback Loop There's been a lot of talk about "good taste" lately. Good taste in design, in hiring, and in product. But what actually produces good taste? Where does it come from, and can it be trained or improved? I wanted to write this because taste feels quite contested right now. With the speed of AI progress, it feels like we've all become beginners in something huge. We're all navigating new tools, new mediums, and new industries that didn't exist five years ago. This leads to a natural anxiety around what remains tasteful, especially when it looks like the majority of our content will be, at least in part, AI-generated. We're anxious, because we can sense we don't yet have the instincts for the things that we are trying to judge. We know we're missing something, but we can't name what it is. Here's a framework I've found useful: Taste emerges from two separate feedback loops working together. The first is what I call the long loop: it is the slow accumulation of experience; of having years of exposure to many variations of something. As you progress through this loop, you build a sense of all the things that are possible in your field. What have people done in the past? What has been tried many times, and yet always failed? What is being worked on now? The long loop is your reference frame. The thing that lets you put anything new that you see into its proper context. The second loop is the short loop: It is the ability to quickly notice small, subtle differences between any two single variations, amongst all the things you've seen. To have an option A and an option B, and to immediately notice that one of them feels slightly off. If you've ever watched a true expert at work, this is one of the first things that is noticeable: *They see way more than you do*. An expert is able to notice things instantly, that a beginner wouldn't have noticed even if given 10x or 20x the amount of time. This level of discernment is not just aesthetic. We see it in all areas where people have attained mastery. In Formula 1 for example, small adjustments to the ride height of the car changes the way air flows under it at high speeds. F1 drivers are famously able to detect changes to the car’s ride height to within **1 millimeter** tolerance. Kelly Slater, one of the great professional surfers, once said that on a perfect wave, he is working with just a few centimeters of error in order to stay "in the pocket". At that level, you are actually measuring just as much as you are surfing. It's for this reason Slater even [goes to great lengths](https://www.newyorker.com/magazine/2018/12/17/kelly-slaters-shock-wave) to create artificial waves with no variable conditions, in order to improve his short loop sensitivity. Ultimately, having good taste requires both loops being well developed. The long loop will give you breadth; the short loop gives you precision. Someone with deep exposure but dull sensitivity has seen everything, but they notice nothing. Someone with sharp discernment but little context is able to spot differences, but they don't know which of these differences actually matter. I've found that thinking about taste in this way makes improving on it surprisingly actionable. > **When your judgment feels unreliable, simply ask: which loop is weak?** > 1. Have I seen enough variations of this thing? > 2. Have I seen a lot, but I need to slow down and train my ability to discern within what I already know? Beginners for example. They always start out lacking the long loop: they haven't seen enough yet. It would be impossible to build sensitivity, or precision, because they don't have the range. Experts have the opposite problem: they've often accumulated so much exposure, that unless they are careful, all of it starts to blend together. The long loop has become strong, but the short one has gone dull. When the long loop becomes too dominant in this way, the brain stops "looking" and starts "predicting." You see what you _expect_ to see based on 20 years of experience, rather than what is actually there. The long loop begins to act as a kind of filter that accidentally blocks out new information. I think this is why so many experienced people have reliable, but quite stale, taste. They've got the pattern recognition, but they've lost the sensitivity. It is also why fields tend to be disrupted by beginners. I think the framework also implies something more general. If taste is simply the combination of two loops, it means that **the path to having good taste is the same across all domains**. Your taste in music won't transfer to design: each domain ultimately needs its own loops. But the structure is always the same. Which means if you've built good taste once, you know exactly how to build it again. And so the anxiety we feel right now, of being beginners again in a landscape that keeps shifting, is really an anxiety about not having the loops yet. But there is reassurance in this: you don't need to be born with good taste. You simply need reps on the long loop, and you need to stay honest about when your short loop is getting lazy. In fact, I think the person with genuinely good taste is the one who keeps both of their loops under a constant tension. Someone who keeps exposing themselves to new variations, while refusing to let that exposure become a shortcut for actually seeing. In the end, I don't think good taste is something one can achieve once and it's done. It's more like staying fit. The moment you stop working at it, it starts to decay. If there's a lesson here, it's that taste isn't a gift. It is a calibrated state, and it requires vigilance. Not just the work of building it, but the ongoing work of noticing when you've stopped noticing.