> I'd bet that 90% of the web only uses 50% of the web standards out there. If implementing 100% of web standards is intractably large and complex, then: Impl

> I'd bet that 90% of the web only uses 50% of the web standards out there. If implementing 100% of web standards is intractably large and complex, then: Implementing 50% of web standards is intractably large and complex too. Unfortunately, you can't turn a humungous problem into a friendly little one by dividing by 2. I would think in terms of dividing by 10 or more. How much of the web only uses 10% of the web standards? My guess is most of it, so we have a chance. > Taking this further: how do we know which features to prioritize? Most front-end devs probably have a rough idea, but what if we got empirical with it? What if we automatically tested the top 10,000 websites or something and made note of which CSS properties they used, which JavaScript APIs they called out to, and ranked them by frequency (and by popularity of the site?). We could chart a clear, direct path toward "what does it take for a browser to be useful in 2020?" This would be great data. Something a bit like caniuse, but showing feature usage rather than browser usage. Getting this data and maintaining it is a full-time job for a team. You can't just fetch the front page of sites to get this info. You need to login and use functionality. Perhaps existing browser telemetry would do a better job, letting users generate the data during normal activity. However you do it, the cost of this idea is mounting up fast :-)