If you put in the page a js that detect the time interval between get and post, it must pass its result as a post parameter.
If I decide to cheat, I can make a post setting the value to anything I like.
This is the fundamental reason against the use of client-side only validation of data: one can always bypass them.
A working strategy could be to assign (server-side) an unique identifier to every get of every page, store it in a db along with a timestamp, and compute the interval (server-side) between the post time and the get stored timestamp. This is certainly possible, but would be an enormous overhead
And, last but not least, cui prodest?
Update: bootnote: I don't think that the idea in itself is good: most of my (few) downvotes are to badly formatted and incomprehensible questions. You don't need (and don't want to spend) much time to decide that a question saying "HELP!!! MY PROGRAM DOESN'T WORK!!!!!" doesn't deserve an answer...
There are a myriad of other possibilities to manipulate javscript - for example you can override some behaviour with costum js (via greaksmonkey), automatically patch it on downloading, fiddling with the js interpreter etc.
Remeber, perlmonks is regularly used by geeks who know very well how to deal with web technology. Just because you can't cheat something doesn't mean it can't be cheated at all.