Per Jessen wrote: > Matt Kettler wrote: > > >>> That's easily checked - we could run a comparison of any up-to-date >>> geoip database against blackholes.us. >>> >>> >> True. >> > > Well, I've answered my own question. I ran a test of maxmind addresses > dated 2007/04/04 against the blackholes.us data dated 2006/05/24. The > maxmind database had 66231 entries/ranges. I took the first of each > one and tested it on blackholes.us. > > Total runtime was 2m5s = 1.9ms per lookup. > > It returned 16300 maxmind entries also found in blackholes.us, but also > 49931 not found. I think the staleness of blackholes.us has been > confirmed. > Again, I'm less interested in staleness than in accuracy, particularly "false-positive" type errors. ie: mis-representing an IP address as belonging in some particular region when it does not.
Really what you've proven here is lack of completeness, which might or might not be due to staleness. (I strongly suspect maxminds database was probably larger and more complete even if you used the 2006/05/24 version of maxmind).. > > /Per Jessen, Zürich > > >
