@Andylew ; @peterdaly and @anyone-who-feels-the-site-operator-works
It works (more or less) depending on what you want to use it for, but when dealing with large sites (or any size of site with more than 1000 pages of indexable AND indexed content) it simply doesn't work.
The -spellingmistake -agkh -[insertRandom] will return different results, and in some cases more correct/accurate results, but it's still very broken.
Your posts got me to reevaluate my stance on the site:- operator and I really gave it my all trying to see if Google might have fixed it after about 12-18 months of documented brokenness.
Like I said, it will do what you want it to most of the time, but if you are using it to see how much of your total website is indexed by Google, it's not going to work.
I tried a site:example.com search for a site I work on. I then tried a bunch of -[random] searches.
I then repeated this for another two sites I work with... and for good measure on two sites I don't work with (but have shared access to).
site:- is still broken for me. I wouldn't trust it as far as I could throw it. I'm not saying it isn't useful. I use it countless times a day. All I'm saying is - as a KPI for your SEO efforts and checking what your indexing is like and all similar intents and purposes - the site:- operator is broken.
Anyone who can show me differently - that would be great!
It used to be one of my favorite (and most powerful tools) for analyzing performance and indexing of deep pages.