Why I love ugly, messy interfaces — and you probably do too
Jonas Downey

Good points, but I also think messy design is a side-effect of poor technology. Not to say that Facebook or Craigslist use poor technology, but that the appropriate technology for these use cases may just not exist yet. Consider Google’s simple interface versus Yahoo’s portal. For a long time you could argue that people loved Yahoo’s portal design — but in reality there just wasn’t a better option. Google came out with Pagerank which opened up the possibility of having a simple interface with just a textbox, and portal websites died.

A messy interface indicates a lack of understanding of your user. Very few users need to see every category on Craigslist’s homepage. But if Craigslist can’t appropriately identify the user then the safest option is to throw the kitchen sink and let the user find the topic in which he/she is interested. In machine learning this would be considered a case of recall versus precision — where recall describes the number of relevant items selected vs. all relevant items and precision describes the number of relevant items selected vs. all selected items.


In the case of a messy interface, recall is a perfect 1. There is probably two or three items of the user is looking for, and congratulations! With your messy interface the target items are there in front of the user’s eyes.

However, precision is something very low, maybe 0.01 or less. Essentially, yes the items are there, but the user is going to have to scour your page to find it.

Going back to the Google example, the power of Google came from raising precision while not sacrificing recall. When they first started, it was very possible that portal websites also had the webpage for which a user was searching, but that page may have been buried somewhere difficult to find. Recall: 1, Precision: microscopic. Google let the user state what they were looking for, then went and found it for them. Recall: usually 1, Precision: probably above 0.5. Every time that Google tweaks its algorithms, it’s 100% aimed to increase both recall and precision.

The clear risk here is that if the technology fails in understanding user needs then recall can be 0 (aka you search on Google and don’t find what you’re looking for). It’s a much safer approach to throw it all on the screen, but one day there will probably be a company who figures out how to give the user what she wants and only what she wants, and that company will win.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.