pointer compression mutation getting outwitted these days

[jchia]

This is how I have been running chinadns 1.3.2-3 for a few years:
chinadns -p 5353 -s 223.5.5.5,1.2.4.8,127.0.0.1:5300 -c /etc/chinadns_chnroute.txt -d -m

It used to be reasonably effective. Lately, I noticed that I often get poisoned DNS query results for typical 404'ed things like google.com and reddit.com.

Maybe the GFW has become smarter against -m. Has anyone noticed this problem, too?

I have a feature suggestion to improve the effectiveness. I have 3 servers, the last one being an honest one outside the GFW: If the first 2 servers (those within the GFW) return results that are in different networks, assume that they are lying and take the result from the last server. I'm guessing from observation that it is inconvenient for the two liars to collude and tell the same lie when they decide to lie.