How I Built a Working Poker Bot, Part 1 - Coding the Wheel
This guy has been writing a series which lays out the framework to create a bot. You really only need a couple years of C experience and good knowledge of whatever OS you code in.
Its interesting because he does not code his bot to win (not that it would be a negative), he codes it so the bot is good enough to break even. With rakeback programs and FPP and the such he can make thousands of dollars a year.
In part two (How I Built a Working Online Poker Bot, Part 2: Interlude - Coding the Wheel (http://www.codingthewheel.com/archives/how-i-built-a-working-online-poker-bot-2)) he gives an example
"If you do the math, you'll find it comes out to around 1.65 million hands per year
. If your bot is playing anything other than the microlimits, this is easily enough to qualify you for (just as an example) the Poker Stars Supernova Elite promotion, which when properly liquidated is worth around seventy or eighty thousand dollars
the year you achieve it, and two or three times that amount the following year (because of the multiplier effect of FPP bonuses)."
So yes this exist and it seems they exist in high numbers. I think as long as there are no colluding swarms (which I also think exist), bots are not too much of a problem since they will just play good hands and try
to classify you as a certain type of player to make decisions.
BTW my first post, glad to be part of the community.