Days Between Updates

Trivial info to while away the time

[atadams]

I took Brett's Google Update History [webmasterworld.com] info and figured out the days between updates.

 Mar 2, 2003 - 36 
Jan 25, 2003 - 24 
 Jan 1, 2003 - 35 
Nov 27, 2002 - 27 
Oct 31, 2002 - 35 
Sep 26, 2002 - 36 
Aug 21, 2002 - 27 
Jul 25, 2002 - 32 
Jun 23, 2002 - 30 
May 24, 2002 - 29 
Apr 25, 2002 - 19 
 Apr 6, 2002 - 45 
Feb 20, 2002 - 26 
Jan 25, 2002 - 29 
Dec 27, 2001 - 32 
Nov 25, 2001 - 28 
Oct 28, 2001 - 42 
Sep 16, 2001 - 28 
Aug 19, 2001 - 31 
Jul 19, 2001 - 27 
Jun 22, 2001 - 32 
May 21, 2001 - 28 
Apr 23, 2001 - 28 
Mar 26, 2001 - 35 
Feb 19, 2001 - 29 
Jan 21, 2001 - 33 
Dec 19, 2000 - 31 
Nov 18, 2000 - 27 
Oct 22, 2000 - 54 
Aug 29, 2000 - 34 
Jul 26, 2000 

• Today (Sunday), we stand at 36 days
  
• There have only been 3 updates > 36 days from the previous
  
• The average is 31.63 (30.75 if you remove the no Sept 2000 update anomaly)
  

Still, we are not overdue for an update because they are never due.

I am working on a chart that correlates the Google update history with Prozac sales

[geckofuel]

Anyone else notice an oscillating pattern that seems to consistently spike (maximum days between updates) around September/October and March/April each year? Perhaps Google makes comparitively major updates every 6 months or so?

The top 5 "days between updates" ended in the following months:

Oct 22, 2000 - 54
Apr 6, 2002 - 45
Oct 28, 2001 - 42
Mar 2, 2003 - 36 (so far)
Sep 26, 2002 - 36

[HuhuFruFru]

haha :-) :-) funny chart, thank you! :)

[Pegasus]

Look on the bright side - we'll get another update in three weeks time :)

What happened when that month went missing? (I wasn't on the scene then.)

[Shak]

atadams = you got it real "Bad" :)

hope the doctor finds the cure very soon.

Shak

[Goanna1]

Thank you for presenting us with factual data rather than groundless speculation.

[atadams]

Perhaps Google makes comparitively major updates every 6 months or so?

That's an interesting observation.

[warmasol]

I think the time between two updates is not so important. For the update google have to compute all the pages of the last deepcrawl. The deepcrawler was late last time: 6th to 16th of feb. The shortest time between deepcrawler stopped and update i saw was 12 days. Now the deepcrawler stopped 14 days before today. I think they are in time and the update will start the next days.

[cornwall]

The average is 31.63 (30.75 if you remove the no Sept 2000 update anomaly)

Perhaps to complete the statistics, someone with time on their hands today could also work out the Standard Deviation!

[Pegasus]

I think it's fairly easy to work out the average time between updates.

It's one month! :D

[Conard]

Just had my whole site deep crawled this morning, for the third time in the last few weeks.
My guess is the update will happen in 24 hours.

[hetzeld]

Perhaps to complete the statistics, someone with time on their hands today could also work out the Standard Deviation!

Average: 31.63
Standard deviation: 6.641

Removing the "no sept 2000 anomaly"
Average: 30.75
Standard Deviation: 5.275

Dan

[yankee]

When Google makes major algo changes, we have to wait longer for the update. A week ago Googleguy mentioned they were working on new scalable spam filters, which mean a longer testing period, etc.

[bobmark]

Scalable Spam Filters?
Will there now be penalty categories running from Naughty to Felony Spamming?

[skipfactor]

One can only hope.

[cornwall]

hetzeld

Thank you for the standard deviation calculation. I knew that someone would have the time to work it out. I always find it useful when looking at means ;)

[quotations]

So in 6 sigma terms, the process is "out of control."

[hetzeld]

Cornwall,

It's pretty easy as std dev is a standard formula in one of the best known spreadsheet program of the planet.
It didn't even take me a full minute to do that as I had a spreadsheet ready with the dates. ;)

FYI, stddev is including the march 2nd dance estimate. If we leave that one off (i.e. limiting to Jan 25th), the figures are: (with and without sept 2000)

Averages 31.48275862 and 30.55555556
Std dev 6.706184191 and 5.272084198

Dan

[Pegasus]

You're actually just determining the average length of a month. :)

[hetzeld]

You're actually just determining the average length of a month. :)

At least this demonstrates that Google has fairly regular update intervals, but I'm not sure employees would appreciate to have their salaries paid with a std.dev. of 6
:))

Dan

[takagi]

> Perhaps Google makes comparitively major updates every 6 months or so?

Perhaps Google wants to put a new number on the homepage?

Aug.(?) 2000
1,060,000,000

Nov.(?) 2000
1,247,340,000

Jan.(?) 2001
1,326,920,000

Oct.(?) 2001
1,610,476,000

Dec 11, 2001 (forum10003/1669.htm)
Google - Searching 2,073,418,204 web pages

Aug 8, 2002 (forum3/4693.htm)
Google - searching 2,469,940,685 web pages

Nov 6, 2002 (forum3/6719.htm)
Google - Searching 3,083,324,652 web pages

[edited by: takagi at 6:44 pm (utc) on Mar. 2, 2003]

[bobmark]

I just ran it through a forecasting model and - according to my parametres - we are overdue as the predicted value was 28.98392 days from January 25, thus February 23.
Useful info, huh :)

[taxpod]

This is the best "Pre" thread I have seen to date. Please do this every month. It is very entertaining at a time when a diversion is needed!

Thanks.

[HuhuFruFru]

yes this is the best pre-update-thread :-)

>Dec 11, 2001 (forum10003/1669.htm)
>Google - Searching 2,073,418,204 web pages

>Aug 8, 2002 (forum3/4693.htm)
>Google - searching 2,469,940,685 web pages

>Nov 6, 2002 (forum3/6719.htm)
>Google - Searching 3,083,324,652 web pages

very interesting, takagi.
so how much can we expect if they really want a new number?
3,500,000,000? 4,000,000,000? :-)

[Jesse_Smith]

Based on these numbers, if Google updated right now, they would not be late!