Quite a bad example in my opinion. Smith is a career 37% three point shooter on over 5 attempts per game, and in a 15 year career (in which he won the 6MOTY award based on his scoring) he shot below 35% over the course of a season 4 times - 2 of those seasons were ones where he played 11 games and 35 games.
Smith is Ray Allen compared to Marcus
Smith was super streaky though. I don't know where to find good stats on shooting consistency, but Smith went through a lot of hot and cold streaks, much as Smart does. I would argue that opponents tended to defend Smith like somebody who was much better than a 37% shooter, and at times he played like it -- I believe he's got some of the top games in NBA history in terms of three point makes.
My point is that opponents have a tendency to respect shooters based on a general perception of the shooters' ability to get hot, not what the shooters' efficiency has been at recently.
Similarly, it's not realistic to expect a player to adjust their shot tendencies over time as their shooting percentages rise and fall. Indeed I think generally you don't want your players to do that. A shooter needs to have the mentality that no matter how many times he's missed, the next one might go in.
Marcus has demonstrated over a large enough sample that he can hit enough threes for it to be a good shot. In fact, he's hit many threes in late game situations, and he's also had plenty of games where he gets really hot from deep.
Yes, you would prefer that he be more consistent, but there is value in a guy who, despite having an overall somewhat low percentage, has the ability to go off from deep. That's not simply statistical randomness. Some players are more consistent or streaky than others. Marcus is a streaky shooter.