Something I've been thinking about lately is the cost/benefits of filtering out bots.
For example on TJ, if you filter by ISPs you have a limit of 20 ISPs you can target, but what if say you have 30 profitable ISPs?
So by cutting out bots in this case we would have to cut out some profitable ISPs and limit our volume.
So my question is how much are bots really costing you?
For example say you have a 10,000 click sample, but 1000 clicks are bots (or 10%). Sounds like a lot right?
But what if these bots have a 100% open and clickthrough rate, that 1000 clicks may cost you only like a $1.
Now I guess the opposite is true if these bots are not at a high CTR and are just eating up your impressions and clicking your banners only occasionally, then this would be costing you a lot of $$$.
I guess my question is is there anyway to calculate what these bots are costing you?
I also think bots have a high CTR, both on the banners and on the LPs ... so the financial loss is not that big as is the ANNOYANCE they are causing by messing with the funnels. I'm afraid we often make bad optimization decisions, because we base it on BOT clicks. That's my main beef with them 
Yup, trying to get rid of the bots directly at the source, by cutting ISPs usually ends up with lower volume and partial collateral damage, when blocking unknown ISPs ... as you stated.
I also share the opinion that networks favor users who don't optimize to much and reward them with higher volume.
Personally, I only cut placements that have too much bots, like a fuckload of them, otherwise I just let it run ... on top of that, on sources like TJ it makes no sense to try to block some sites since they are all legit.
What would make sense to me, but I wasn't able to solve this in a foolproof way yet, would be to cut the bots at the tracker and just send them elsewhere, so they don't mess with the optimization.