I've been reading Satoshi's paper recently. There's one thing I cannot think through.

That is because the target will limit the block production speed to 2016 blocks per two weeks on average no matter how high the hashrate gets. What if the attacker forks the blockchain from height 1 with a constant difficulty 1 and produces exactly 2016 blocks per two weeks (just in timestamps, not in real time). I think it will be very easy with such low difficulty to catch up current blockchain's length, and eventually surpass. According to the longest blockchain concensus, will the fake blockchain replace the current one?

I would be grateful if someone could clear up my confusion.