Ori, I don't think you addressed the point I made about that study. They didn't ask users what they thought *their* browser setting meant and what they expected. They asked what they thought a big red button with "DO NOT TRACK" on it meant -- and the most common answer had to do with their local browser history!
Regardless, I think you make a good point. The cost of getting something wrong here may not be symmetrical, but it's not clear to me that erring on collecting absolutely no data is less costly.
For example, not collecting usage data about certain sections of our population (e.g. IE10 users where DNT is set by default) means that we don't know if our software works for them. This isn't free, and in the long-term, it can have substantial negative effects. If DNT was always disabled by default in major browsers, I would expect such biases to be minimal.
Also, I think that if a user sets DNT and expects it to do something it isn't supposed to do, we can always point them to the spec. It's a sad fact that, if you want to remain private on the web, you're going to need to inform yourself about how such things work. Just because we adopt an extreme/overly-simplistic doesn't mean that the people you really don't want to have your behavioral data will to -- but it certainly has the potential to make research & product's job much more difficult.
Really, what I'm trying to say is that if I "decline to collect data about [you]", you shouldn't say, "meh". You should be concerned about how we're not considering what works and does not work for people like you when we design, test and deploy software changes. In a way, it's like taking away your vote. And if you don't believe that, I'd like to suggest that the only alternative is that the work that I do does not bring value to our users -- and I'd beg to differ.