Use a diamond pipe and place one item in each output you want them to use. It will distribute them evenly that way, I believe.
I don't see why a diamond pipe filter that's not filtering would distribute any better than a normal pipe.
My understanding is that, unless a pipe has special rules, items will choose an exit at random, with the restriction that they won't go back out the direction they entered. Wood, iron, and diamond all have special rules. For diamond pipes, if an item is in the filters at all, the direction is again chosen randomly among each direction that has a matching filter. And this can be biased by putting multiple copies of the item type into a filter. That's all out of the documentation.
I said to place a diamond pipe there and put one item into each output color. Then technically it IS filtering. It sees, for example, one glowstone on red, blue and black. It sends glowstone to those outputs evenly. If you put 2 glowstone on red and then 1 black and 1 blue, it would send 50% of the glowstone to red and 25% to black and blue respectively.
It's random. Random. It will average even over time, unless you bias it (as we both mentioned). But it doesn't remember where it sent the last item. So it is not guaranteed to split evenly. And is therefore no better than a dirt-cheap cobblestone pipe.
Okay, calm down. I did indicate that I wasn't sure, but I thought that if it was actively filtering then it would distribute it evenly and intelligently. Good to know.
well i hope we all remember from computer science class that computers cannot truly generate random numbers, but they can simulate it so it appears random to us. It an item can go one of only two directions down a pipe, it will pick at "random" each time (durrr, i know right?) with no regard to what has come before (again, random). In the short term, even over a real day it can certainly develop a lopsided distribution. Go ahead and flip a coin for 1 hour and see what happens. Since the coin has no memory, any lopsided distribution of head or tails may or may in fact be exaggerated in the long run.
The more times you flip, the GREATER the difference, but at the same time, the more you flip the closer the PERCENTAGE will be to 50-50!
Example:
I flip a coin 100 times. Tails come up 60 and heads come up 40. The difference is 20 and the percentage is 60-40.
Now I flip it 1,000 times. NOW the difference might easily be off by more than 20... say it's 550-450. The difference in flips is 100 but NOW the percentage is closer... 55-45!
Again, the more times you flip, the percentage will always get closer and closer to 50-50, but the difference will also be off by a greater amount!
random does NOT mean it will average over time. Sorry.