Consider that every page that the crawler must download is a page's worth of bandwidth your server must deliver. If these pages are generated, your server need to spend the computation to generate that too.
Add to this the fact that a crawler is able to stop and back off if its resource constraints are being overwhelmed, but as a webserver that may be serving content that is "mission-critical", for lack of a better phrase, your server probably needs to maintain a low latency. In addition, there are multiple web crawlers out there, that could hit your server at the same time.
Given that, I think the scheme you've described is a mug's game, where for every dollar's worth of damage you do to a crawler, you're doing more than a dollar of damage to yourself.
It might be possible to kill a naive crawler with some infinitely-spiralling link web. However, most modern crawlers are built with this in mind and will stop after a certain recursive level is met.
Add to this the fact that a crawler is able to stop and back off if its resource constraints are being overwhelmed, but as a webserver that may be serving content that is "mission-critical", for lack of a better phrase, your server probably needs to maintain a low latency. In addition, there are multiple web crawlers out there, that could hit your server at the same time.
Given that, I think the scheme you've described is a mug's game, where for every dollar's worth of damage you do to a crawler, you're doing more than a dollar of damage to yourself.