Trust the vibes
Great episode of the Shop Talk Show here with Chris and Dave where they chatted about that whole vibe driven development thing and I worried listening back to the episode that my original post was...misleading. And arrogant. Okay, and pretty dismissive, too.
So let me clarify.
When I think of vibe-driven development I don’t mean that everyone in an organization should just do whatever the hell it is they want to do. That’s pure chaos. Instead, by using that phrase, I want to look critically at how decisions get made when making software—and point at the enormous failures and missed opportunities of tech companies I’ve seen over and over again where folks lean on customer feedback and junk data to make decisions.
Data is often used to patch up a lack of confidence in the decision making process. It’s like an anxiety blanket. If everyone in a room can stare at a chart and say, ah yes, this!, then an ethereal, intangible problem transforms and becomes solid, physical, sensible. The problem becomes rooted in the clean laws of physics and spreadsheets instead of the incomprehensible and often sticky realm of human people.
But do we really need data to tell us that adding this giant modal that pops up and interrupts the flow of work is bad? Sure, the data says it makes a point in a spreadsheet go up by 0.1% but how do you feel about that? Is it disrespectful? Does it suck? Is it good for us in the long run, for people to be annoyed by our product? Would you be ok making that thing for your family, your loved ones? Then why is it okay to show this kind of junk to a customer? Trust your vibes, man.
Sometimes data and customer feedback can point to important problems! I totally agree with Chris and Dave on this front and I’m not advocating for selfishness here at all. Vibes-driven development doesn’t mean ignoring data entirely. For example, data is handy when a lot of people bounce when they hit a complicated flow or when they get stuck or lost somewhere in your app. This sort of data is like a canary in a coal mine—it can sometimes identify problems that you’ve been ignoring but but but (and this is the important bit) data cannot tell you how to deal with those problems!
Data can’t tell you shit about typography or beautiful illustrations. It can’t tell you anything particularly useful about design systems or what to refactor next. (Side rant: when folks at tech companies talk about data they’re not talking about a well-researched study from a lab but actually wildly inconsistent and untrustworthy data scraped from an analytics dashboard.) These numbers—any numbers, it doesn’t matter how they’re collected—are not representative of human experience or human behavior and can’t tell you anything about beauty or harmony or how to be funny or what to do next and then how to do it.
The only way to figure out anything meaningful is by prioritizing your use of the product over everything else and then trusting yourself (or your employees!) to make decisions based primarily on gut feelings and experience: Is this cool? Is this kind?, etc. etc.