Notes · 15 August 2019
Browser support is no longer the conversation
We spent a decade arguing about IE. The argument has moved.
We spent the last decade arguing about which browsers a site should support. The argument is over.
For most of the small-studio web, the answer is now: the last two versions of Chrome, Firefox, Safari, and Edge, plus the current iOS Safari. That is the support matrix. It covers more than 95% of users on any audience that is not a government tender, and it lets the designer use almost everything in the modern CSS spec without the polyfill conversation.
Internet Explorer 11 is no longer in the support matrix. It has not been, for any new project, in over two years. Anyone telling you it should be has not run the numbers on their actual analytics. (The exception, government and large-enterprise procurement, is real. It is also not the small-studio web.)
The conversation that has replaced “which browsers do we support” is a more useful one: which subset of modern CSS does this team feel comfortable using.
A site that uses CSS Grid, Flexbox, custom properties, and clamp() is uncontroversial in 2019. A site that uses Container Queries, the :has() selector, and oklch() colour is on the edge. A site that uses View Transitions or scroll-driven animations is a year ahead of where most teams are. The decision is not “which browsers can run this”, because the answer for almost everything is “all of them”. The decision is “which features are stable enough that the team can rely on them for the next two years”.
For most projects, my default is now: use anything in Baseline that is at least one year old, treat anything newer than that as progressive enhancement, and stop worrying about the rest.
The tools that have made this easier: Caniuse for individual features, Baseline for grouped support status, and Browserslist for codifying the matrix in the build pipeline. None of these are new. Most small studios are still not using them as the basis of the conversation.
The work has moved on. The argument has not. Half the front-end articles I read in 2019 are still arguing about IE11. Stop. Move the conversation.