I wouldn't say Google error messages are that rare. Well, on the search engine but not on GMail, etc.

I agree, you do get a lot of those "oops, connection failed" things, especially on Gmail and Orkut. Maybe those are plain errors in disguise. :) But I think it's rare you get a screen reading "error" or something. I recently had a real live error on a Google web search and it really stood out because it's so rare...

Error messages are not so rare yes, but NOT with those parameters in them ..

You usually get an encoded hash code..

Great find Philipp.. *going to dig the info*

Very interesting.. this seems to be a buffer overflow leak!!

What could have caused that ??

Are this PR algorithms?

> Great find Philipp..

I'm forwarding the thanks to Mathias Schindler who found this info nugget (to use PD's terminology) :)

Thanks to him too then ..

This sounds like some of the variables that the script uses to format the SERPs.

For instance, the line:

--sections_to_retrieve=body+url+compactanchors

My only other idea is that when I'm having trouble with code I put small bits of code in to output status messages and print out variables along the way, until I find something that didn't work. Maybe we caught them working? :)

Demen

Good fake.

Sorry for the double post, but I had another idea. The top half or so without the double dash "--" looks like a listing of the current server state of one of Google's document servers! And also a history of all of the queries on it (The queries_total 1229799558, for instance, as well as the 2800000000 i think being a display of a 2.8 GHz processor. Then the num-docinfo-disk_total looks like the number of cached and indexed documents on that server). This all leads me to believe they got into the diagnostice mode of the server, and not neccessarily an error message...If you don't count unexpected diagnostics as errors.

Demen

A more technical friend of mine gave some high-tech explanation on my blog ..

Maybe some of you can figure it out ?

http://www.seopedia.org/seo-experiments/google-error-reveals-secret-details/

I think he's saying (Ciprian) that the top half is just server information and the bottom half is "arguments" (in this case, values passed to a computer program when it starts) that activate modules (other pieces of code that together run google) in the program. Which leads me to believe that the server was restarting or had somehow been reset, as I don't get an error message when I try the cache query. Possibly a technician or maybe an automatic system had restarted the "Google Search Program" (now I'm making stuff up!) and we found them at that exact time. That would be incredible timing, but possible.

One last thing- right behind the expiremental spamscore that everyone's talking about is another expiremental value with a "no" in front of it!! Indy Rank. Could this be a new ranking system? Or am I missing something?

Demen

>> That would be incredible timing

Not really.. Imagine de hundreds of thousand of users using Google each minute.

No this cant be a fake :)-

its a classic wrapper on The Gimp Toolkit, or GTK+ type of variables.. Secondly port 32311 seems a likely freebsd type of mt kernal

anyone else has ideas to throw around ??

timedout-queries_total 14227
queries_total 1229799558

This means: 0.00115% of the queries are timed out. Not bad. Not bad at all ;)

Actually I would consider that error margin pretty bad ..

0,001 isn't that small ..

That means that 1 in a 1000 users, gets an error .. It's freekin' BAD.

Hey

Actually i did the initial find, and i was the first one to post it at the DP forums, so i think you better thank ms :D

http://www.abuzant.com/od/2006-07/google-server-errors-fact-or-fiction.html
http://forums.digitalpoint.com/showthread.php?p=1073136

Yes but he said 0.001 PERCENT which is one in one hundred thousand

To confirm what some people have already said above, the lines starting with dash-dash on the bottom half are program options. The dash-dash is a common convention for program options with modern UNIX-type systems (Linux or FreeBSD).

And I concur the top half is server statistics.

>>>> 0,001 isn't that small ..

>>>> That means that 1 in a 1000 users,

Actually is one a 100.000 users. :)

yes it is : 0,001 per cent = 1 per one hundred thousand

If anyone is ever missing Google errors, here is how to reproduce one:

http://www.google.com/sorry/misc/?continue=

Presto!

huh ?? you got a 403 fobidden error :)-

Matt confirms this error message is real...

"Yes, it was real. No, I’m not going to comment on what any of it means. :)"
http://www.mattcutts.com/blog/catching-up-2/

Some thoughts as to what the different parameters mean:

http://ofb.net/~wtanaka/node/221