It's all about balance. Water, fluid and ever moving wears away the stone...
Sorry, I blacked out for a second. Anyway, what you REALLY wanna do is figure out which queries can get you your result set the fastest.
Sometimes one big query is way faster than a bunch of little ones, sometimes not.
Build your query, then echo it to the screen. Then cut and paste it into whatever tool you databases uses for direct human interaction and explain / analyze it. If you've got a choice between one big honker of a query that runs in 400 ms, and a dozen that each run in 5 ms, then the dozen win.
but maybe over time, as the database is populated, the big honker will stay at 400ms while the little ones will rise to 20, 30, even 50 ms as the database gets bigger and individual queries result in more random I/O than before.
It's not how many queries, and it's not how much that they "cost" in CPU/ I/O. It's how many times how much they cost that counts. The product of the two.