Edward B. Dreger wrote:
Apologies for forking yet another thread from one which I myself have been largely ignoring. AFAICT, though, most posts have shown little interest in combining different approaches:
* Provide a "default" sandbox.
* Allow unrestricted access -- perhaps after a quiz, perhaps when a user activates a form.
* Let IDS trigger sandbox mode.
* Provide IDS-triggered-sandbox override for those who agree in writing to <blah>.
* Anyone in the IDS-proof class who spews filth deserves to be fined.
* Use different IP ranges for different service classes. Flag in rwhois, a special RRTYPE, or whatever suits one's fancy. (This assumes that providers could agree on a standard.)
Perhaps no one tactic fixes everything. Fine. I readily admit that the above combination isn't a miracle cure. But is there a moderate chance for improvement? I think so.
*applause* Whilst I may not communicate it sometimes - this sums up all my thoughts on the matter quite neatly, you have my whole hearted support. / Mat