It's because all cryptography was done (almost) by hand and most importantly the people trying to break it were only really using metal methods. Nowadays people use computers to break your codes, so your code has to be much more complicated. The only way to do that is with a computer.
Sounds like a pretty weak cypher by todays standards. 90 million combinations seems like it would be fairly easy to brute force. Still, it's interesting to see that some of his ideas, like row operations on matrices, are still used in some crypto algos today.
Indeed, by todays standards it's pretty weak. Then again, 200 years is a long time, and back in the day all the grunt work had to be performed by humans instead of computers. Brute forcing was not a realistic possibility.
Correct me if I'm wrong, but it almost sounds like a rudimentary AES-type encryption when they describe how he broke it into a grid and then jumbled the lines. I'm no cryptographer, but the description of this compared to AES encryption seemed remarkably close.
That said, using digraph probability to solve it was inspired.