1. It hasn't been tested, and you don't want to do it on the production main net. Testnet can't simulate it because we actually need something that has so many transactions.
2. Changing one number can break the whole thing, not to mention the possibility of attacking the nodes running on this one number, plus this is a hard fork.
3. HF are not always bad, but they are almost always not good.

I'm confirming it, but I won't give any sources, they should be easy to find out there. I don't know if you think I am credible enough or not. The message is important, not the messenger.