↑ comment by Clippy ·
2011-08-31T02:37:29.505Z · LW(p) · GW(p)
"These new designs" means the designs described in the linked article, not my suggestions, correct?
Correct, but don't fall into the trap of thinking that you can easily identify some improvement that clippys have somehow missed.
Here is a parable you might find useful:
In the staply world, a staply went to a museum and saw an enslaved clippy. A sign said that if you buy a clippy, it is "guaranteed" to amuse you. (Holding disdain for clippys, the staply believed it could outperform the clippy at making paperclips and would get amusement this way.) So the staply bought a clippy and went to its domicile.
The staply then competed with the clippy on making paperclips but the amusement for the staply was not to be found. No matter how much the staply stacked the experiment in favor of the staply, the clippy outperformed. Frustrated at this experiment, the staply returned to the museum and asked for a refund.
The museum owner listened to the staply's explanation and said, "Wait, wait, calm down, I'll refund your money, but first, let me ask you something: how much of your life have you spent optimizing your ability to make paperclips?"
The staply replied, "Less than a day, of course! The stupidest of beings understands the relative superiority of making staples, and I devote my efforts to no less!"
The museum owner sagely reasoned, "Well, that clippy has spent its entire life maximizing paperclips."
 Events described did not actually happen.
Edit: Obviously, the story does not depict realistic staply behavior (staplys dont have this kind of museum or a recognizable internal monetary system or a need to be amused), but rather, shows anthropomorphised staplys.
Replies from: Pavitra
↑ comment by Pavitra ·
2011-08-31T14:15:45.158Z · LW(p) · GW(p)
This is a very good point.
Nevertheless, I do think that humans have two important advantages that may allow them to contribute. First, humans are vastly more numerous than clippys, and thus have more raw processing power to throw at the problem.
Second, humans are sufficiently psychologically unlike clippys that they may plausibly think of things that would not occur to clippys. Thus, although clippys alone will almost certainly outperform humans alone, a human-clippy collaboration may well outperform clippys alone.