I see, your right that seeing a string of zeros denoting a bunch of alt currency/BTC pairs isn't pretty.
On the other hand though, if you do add almost any other alt coin, the exchange rates will be a very large numbers.
It's not an issue with how pretty the numbers look.
We should walk through this with an example to make it really clear what the issue is. Suppose we have an exchange that, due to resource limitations, is limited to 5 decimal places for orders. This exchange wants to add a currency pair so that users can trade between currencies x and y. Should the exchange use pair x/y or y/x?
Let's say that given current market prices on other exchanges, the price of x/y is currently around 0.01. But the exchange wants to allow for the possibility that this price could crash by at least a factor of 100. After all, currency x isn't nearly as well established as currency y and it seems much more likely that currency x might crash relative to currency y than the other way round. So, the exchange wants to allow for the possibility that x/y price might drop to a price of 0.0001.
But if that happens, the exchange will only have one decimal place left to work with. Since the exchange is limited to 5 decimal places, it can only allow traders to place orders in increments of 0.00001. But there's only 10 such increments between 0 and 0.0001. This means that if price of x/y dropped to 0.0001, trading would be *severely* limited. To get a sense for how limited it would be, suppose that we limited the order book for XBT/EUR to increments of 50 EUR. So you could place a limit buy or sell for 500 or 550, but nothing in between. It would be mighty inconvenient already to have orders limited to increments of 5 or 1, but 50 would be a disaster - liquidity would be absolutely horrible.
This problem disappears if the exchange goes with pair y/x instead. The price of pair y/x is currently around 100. If currency x crashes relative to y by a factor of 100, then the price of y/x will go up to 10,000 (congrats y/x longs!
). But this doesn't present any problem for trading on the exchange - orders can still be placed with very fine-grained precision.
Hence, given the scenario the exchange wants to be prepared for (currency x crashing relative to y), pair y/x is the better choice. Now if all the other exchanges are using pair x/y, this makes the choice less clear-cut, but still it's not unreasonable to think that y/x is the better choice.
The above is essentially the reasoning that led us to select XBT/LTC rather than LTC/XBT. We're not LTC bears, but we do figure that LTC is more likely to crash relative to XBT than the other way round.
Why don't you write 26.32m for 0.02632 ?
That's just a display trick. We'd still have to keep track of things computationally, which is the real issue. One thing that might work though is to offer the pair LTC/mXBT, where mXBT = 0.001 XBT. I think that'd be the way to implement your basic idea. But I'm not suggesting we'd do this.
Edit: adding text below for easier search.
Reason for inverted pairs on Kraken
... I hope this won't happen again for a long time! 
