What is the most up-to-date, robust, and reliable way to check verb (or other POS) frequencies in current usage?
Is there any hope of an algorithm involving counting Google hits and dividing by some benchmark number (say 1,000,000,000,000) or is that so error-prone as to be out of the question?
(Any help with tags would be appreciated.)
Use a professionally created corpus, like the Corpus of Contemporary American English. You get much more reliable data and examples, plus you can do sophisticated part-of-speech searches.
A better way is to find a word used in texts. Most language experts analyze books, newspapers and magazines that are edited before they are published rather than taking on the Internet as a whole. You can choose what countries and target demographics by this method as well. I would go for quality over quanity otherwise your numbers are meaningless. The word "friend" is frequently misspelled "freind". It is a common error, but not standard English.
