While the idea of another program to serve DNS isn't all that bad, I think jumping ship just because of one new policy isn't necessarily the most prudent thing to do.
The new policy may not be the only reason; the bugs in BIND 4/BIND 8 are making everyone consider what to use as replacement: BIND 9, djbdns or something else. Both BIND 9 and djbdns have non-technical issues; BIND 9 licensing is good, but ISC sticks to security notification methods that are not. Licensing is a djbdns weakness.
WRT djbdns: I've had a moderate level of experience with it, and, while it seems interesting to an extent, operationally I've had several annoying encounters with it.
When challenged, I seem to get the reply of "maybe some time later it will have that" or "that is insecure, djb doesn't support that".
What operational issues are annoying and in what daemons (dnscache, tiny-dns, axfr-dns, wall-dns) ? Needs like authoritative servers and recursive resolvers are different, and may be a djbdns/BIND9 mix can perform better.
djbdns is also very infant - it's probably not popular enough for all the skr1pt k1dd13s to have an interest in hacking at, because finding a vulnerability in djbdns is about as useful to the "wreaker or havoc" as finding a master door and ignition key to a '58 pinto -- there's about 17 of them on the planet :-)
djb himself seems not to be very popular; I bet that are many people out there trying to find bugs in his software just to make him look silly. Rubens Kuhl Jr.