I have also been thinking about this and I agree. Seeing that upvote button just makes it too tempting to click on in the feed if it looks like something that will be voted up.
It is simple psychology that we are driven from a biological perspective to take cognitive shortcuts when we can. Having the up-vote visible on the feed plays to that and is hard to resist because it saves us time and mental energy.
Removing it from the feed will reduce the likelihood but obviously won't eliminate it. It would be interesting to collect data on this (assuming it were possible) to see how many posts get upvoted without being opened.
An even better data point to collect would be to see how quickly those that open posts then go on to vote on the same post. How does that time fit with the time it would take an average person to read the post? I'm not sure how feasible this kind of information would be to collect as I'm not a programmer.
If it were possible to collect this data it might be useful to do a before and after comparison to see how many people were actually reading the post before and after removing the upvote button on feeds.
It is very tempting to click! I find myself doing it unconsciously sometimes and going back on it seems risky since it seems the world watches your every move. ognitive shortcuts is exactly right.
Data collection wise I don't think an upvote has a typefield where it contains the method of upvoting but maybe it should? I would be very interested to see this data.
I actually haven't ever upvoted anything without at least clicking in and looking at the post. But that isn't to say that I haven't skimmed post and upvoted them in a matter of 5 seconds. Not long enough to actually read the entire thing. Steemit is an interesting experiment in rewards. With the follow button it becomes a question of do I follow users that I like but aren't upvoted by whales or do I follow people that I don't care about their content and just upvote them because I know the whales will. It becomes an issue. Great post