1. How is this different to Glassdoor?
2. How do you get enough data points to be useful?
3. How do you account for selection bias?
4., (and here's the big one) how do you account for the mistruths of self-reported income?
The last one is the big problem with all self-reported income summaries. One thing I've learned over the year is that people lie, even anonymously. I'm not even sure why. Do they really believe it? Are they posting it to make themselves feel good? I really don't know. But I do know that I don't believe your salary claim until I see a W2.
As for selection bias, this is a big one too. Take a company like Google. At level 7+ data point get pretty thin. This is partially because there aren't that many but the real problem is that these people really have no motivation to share. So you don't really see the big outliers.
Lastly, most of these aren't clear on what number they are report and what users should report. Is it base salary? What about bonus? What about RSUs/options?
As anyone who has worked in big tech knows, base salary number are almost completely useless. Salary band is defined by level (pretty tightly, in my experience). More than 50% of your total compensation will come from bonus and stock and stock in particular has massive variance.