There is a huge market for e-mail harvesting and spamming software. Literally hundreds of companies have their own custom software applications designed for spamming and harvesting. Already, harvesting bots search hundreds of sites at once, and send spam while harvesting addresses.
In order to remain competitive, companies that make harvesting software need to make their products better able to capture e-mail addresses from websites. This means making the bots harder to detect and more difficult to block, in addition to making them better at finding hidden e-mail addresses.
Harvester bots have been designed to emulate human behaviour. Already, harvester bots pretend that they are using a web browser to look through sites. Rudimentary language processing will allow spam harvester bots to defeat simple obfuscation techniques such as user[at]domain[dot]com, and avoid traps like "do *not* send e-mail to user@domain.com".
A common method of protecting an e-mail address is to use javascript code to mask the HTML that displays an e-mail address. Harvester bots can already crack simple javascript masking, and it won't be long until harvester bots can get around any javascript masking technique.
Some people replace their e-mail address with a picture of text. Since this hides the HTML code that shows the address, it offers fairly good protection. However, newer harvester bots are equipped with OCR (optical character recognition) software that analyzes pictures and picks out the e-mail address.