But I think the opposite case still kind of holds, that with P/E you can at least see if a company is most likely not overvalued. And that was the issue in question ("everything is just gambling").
There are still measures that correlate well with low risk and strong returns for some subset of companies, but identifying a subset and building valuation models for them is non-trivial (e.g. I typically use risk models for revenue growth in comparative valuation which don't even apply to most of the market). If it was as simple as looking at a trivial ratio of public numbers, everyone would already be doing it.
I've been investing a long time and the markets have changed a lot over the decades. At this point, I think most of the investing advice from several decades ago is obsolete because it is based on assumptions that aren't actually true today. Investment advice and heuristics have a shelf-life. Most people aren't going to build a portfolio strategy from first principles, it is a lot of work, hence the popularity of index funds.