personally i like to keep scoring systems simple and transparent. speaking very generally, i would prefer thresholds and flat, binary bonuses over variable and scaling bonuses. in other words, i would rather add a 'lengthy capture' bonus than make the current capture bonus more complicated (hypothetically speaking).
point captured: +30
point man: +15 (if personal channel time for capture exceeds the channel time of teammates)
lengthy capture: +15 (if personal channel time for capture exceeds a static value)
point captured: min +30, max +60 (based on contribution and capture time)
the simple example works to achieve the same goals as the complex example, but without as many unknowns to the player. a new player wouldn't necessarily know why they got 34 points in one case and 57 in another. in the simple example, they would learn (relatively) quickly that capturing is important, but aiding to capture is less valuable when most of the work is already done and not much time would be shaved off with their assistance. the tradeoff is that the simple version is less accurate but i think it's a worthwhile trade if it means that scoring will achieve more than being a measurement of contribution.
base points + bonus points * (time captured * modifer)
Thus the longer you help the more points you get, thats pretty easy to figure out.
Another issue that isn't addressed is how poor performance by the "bot lane" player can yield extremely high scores. Ex. He constantly loses bot tower over and over but keeps managing to take it back each time. He will have a high score and a high capture rate leading him to believe he did well when in fact he is probably the reason for the loss of the game.
Honestly the biggest thing I am worried about is when ranked dominion comes around if "Score" is still used as it is today, it will be a huge deterrent to good gameplay.