Rounding Off a Large Number of Bits Is Not as Secure as Adding Error

The idea with rounding is, like, you've got some gladis point and then you're supposed to add some error. Most of the schemes will add some error by just choosing random error from a small interval. The alternative idea is that if you have the lattest point, and you just roundet off to the nearest multiple of ten. This makes things smaller. Instead of having to add error and send the whole thingo ho, just rounded all and you kind of just dropped the lott bits. So, you know, why is this secure? I don't know. Aso it migh now, as faras, as far as we know, i mean, if i