Web Platform Tests

For the internet, there are fixed rules that precisely define how a webpage is structured and how it should behave. These rules are decided by international bodies and exist online as lengthy technical documents. Programs like Chrome, Firefox, or Safari must adhere to them, otherwise the same page would look different everywhere. Web Platform Tests are a huge collection of small test tasks that check exactly this. Each task poses a tiny question, such as: Is this text really centered? And every program that displays webpages can run through these tasks and receives a score at the end.

Why the same page should look the same everywhere

In the 2000s, the web was a patchwork. Many sites carried the notice “Optimized for Internet Explorer” because they looked broken in other programs. Companies had to build their websites multiple times, a separate version for each display program. This cost money and held back small providers who couldn’t afford this effort.

Web Platform Tests tackle this problem at its root. Because all manufacturers use the same test collection, there is a common standard. A manufacturer can no longer claim that their interpretation of the rules is just as valid. The test simply says: passed or not passed.

There is also an economic side to this. The more uniformly the programs function, the harder it becomes for a single corporation to lock in the web for itself. Competition authorities therefore pay close attention when a manufacturer deliberately ignores important rules. Test results provide hard, verifiable numbers instead of opinions.

A test collection that all manufacturers help write

The tests are kept in a public archive that anyone can view. Employees from Google, Mozilla, Apple, and Microsoft write new tasks there together, as do volunteers. In total, there are over a million individual tests. Anyone proposing a new rule for the web today usually submits matching tests directly along with it.

Technically, a task consists of a tiny webpage. It creates a specific situation and then checks itself whether the result is correct. For visible things like layout, a reference image is used: the test page must look exactly like a second page that produces the same result in a simple way. If both images match, the test is considered passed.

The results come together on a public overview page. There, for each test, it states which program passes it and which does not. Particular attention is paid to the number of tests that work simultaneously across all major programs. A common misconception, by the way, is that a high score automatically means a good browser. It only measures rule compliance, not speed, privacy, or ease of use.

From Interop rankings to your own school website

The tests become most visible during the annual “Interop” initiative. In it, manufacturers agree on a list of areas that still don’t work well together. Progress is then measured via Web Platform Tests and displayed in a public ranking. Tech news sites regularly report on these figures.

Even without programming knowledge, you benefit from this daily. That a video player, an online form, or a maps application works the same on a phone and on a laptop is not something to be taken for granted. Thousands of passed tests stand behind it. Anyone building a small website themselves, say for a school project, therefore has to try it out individually in every program far less often today.

Web Platform Tests should not be confused with testing tools for individual websites. Programs like Selenium or Playwright check whether a particular online shop works correctly. Web Platform Tests, on the other hand, check the display program itself. So one checks the car, the other checks the road that all cars drive on.