Why Playwright sucks for end-to-end tests in 2025?

There are so many ways to get locators wrong.

I have been implementing self correcting locators in my automation frameworks since QTP. The fact is the longer an automation project goes, the bigger that page object model becomes. The bigger the model, the more maintenance when things change. Maintenance is a ROI killer.

This article makes a good point that identifying an element every time it's seen on the page like it's the first time is great. No more looking for an anchor class element when it changed to since the last Sprint. And it usually is just one of many failing my test suite.

This looks cool.