I blogged a while ago about strings hidden inside the various Google Talk executable and what could be inferred from them, and it looks like more Google internals have been revealed by the following logging information from a Google error page on a cache query:

pacemaker-alarm-delay-in-ms-overall-sum 2341989
pacemaker-alarm-delay-in-ms-total-count 7776761
cpu-utilization 1.28
cpu-speed 2800000000
timedout-queries_total 14227
num-docinfo_total 10680907
avg-latency-ms_total 3545152552
num-docinfo_total 10680907
num-docinfo-disk_total 2200918
queries_total 1229799558
e_supplemental=150000 –pagerank_cutoff_decrease_per_round=100 –pagerank_cutoff_increase_per_round=500 –parents=12,13,14,15,16,17,18,19,20,21,22,23 –pass_country_to_leaves –phil_max_doc_activation=0.5 –port_base=32311 –production –rewrite_noncompositional_compounds –rpc_resolve_unreachable_servers –scale_prvec4_to_prvec –sections_to_retrieve=body+url+compactanchors –servlets=ascorer –supplemental_tier_section=body+url+compactanchors –threaded_logging –nouse_compressed_urls –use_domain_match –nouse_experimental_indyrank –use_experimental_spamscore –use_gwd –use_query_classifier –use_spamscore –using_borg

It’s hard to interpret what exactly these codes mean, except that they give “cooking” parameters for pagerank computation to establish limits on the convergance of pagerank over time, that the server’s a 2.8Ghz machine, and a bunch of timing information. Thanks to the parameters section, though, we can try to infer:

• Country of Origin and/or language are interpreted per-domain and passed onto pages in the domain tree
• Production code is being used
• Pages are parsed into discrete units of meaning, such as sentences
• Some product of vectors optimization is being done (to aid in PR computation, etc)
• The body, your links, and the url are the most interesting to Google
• The bord are involved!!

We’ll probably only learn more about the internals of Google over time by slowly watching how their product works. Any suggestions, leave comment?

blog comments powered by Disqus