About Google update in General

How do they start the update

[zeus]

I sometimes wonder if they just push a button or they got a program that starts at a special time late every month or maybe they have a meeting where they talk about the best day to start to update Google.

zeus

[progen]

I'd go with a meeting. But who knows.

[Andrew Thomas]

If it is a button, i hope they would hurry up and push it!

Or maybe it is stuck :)

[ukgimp]

I heard, rightly or wrongly that it was planetary alignment.

Perhaps GoogleGuy can confirm or deny that.

[Nick_W]

Eh? What are you lot gabbling about ;)

Everyone knows they wait untill there have been 100 false postings on WebmasterWorld!

(counting 59 at the moment, reckon thurs/fri should do it...)

Nick

[bateman_ap]

It is prob a set date of the month unless they aren't happy with the changes they have made with the algorithm and push it back a day or so. i imagine there are many internal servers on constant update there showing them a smaller set of results to fine tweak settings.

[edit_g]

"i imagine there are many internal servers on constant update there showing them a smaller set of results to fine tweak settings."

Yup, I second that.

[vitaplease]

They have five PhD programming brigades.

Each have to solve some algo improvement from suggestions from the "Friday happy-hour-feel-free-to-mention anything" evening and from the top-three most irritant spam-report list.

Full moon is the deadline.

Failing to finish by full moon means next month you get to answer PR0-emails or you get to scan catalogs.
(this month many email will get answered ;)).

Googleguy posting at WebmasterWorld at two AM local time in the morning is a good sign the deadline is aproaching or has been surpassed..

[zeus]

Well there is some interesting comments here, they do not go by the calendar, full moon, I dont even think we could find out the math. way, maybe it is a meeting thing.

zeus

[Sinner_G]

Maybe they just drink beer the whole month while Googlebot is doing the hard work and then press the button (probably the enter button) when all the beer is gone?

[Weblamer]

They have a secret ceremony in the basement of the google building. There they dress in black robes and chant the googlebot chant, and in the center of the room there is an altar with a big red button. Once the chanting begins to peak the high googlebot preist cries 'LET THE UPDATE BEGIN!' and slaps his hand on the big button on the altar.

At least, that's how i picture it.

[ppg]

You missed out the bit where they sacrifice a few screaming webmasters to googlebot chanting "PageRank 0, PageRank 0" ....

[rfgdxm1]

Very credible source confirm they do sacrifice webmasters. However, they must be virgins, which means I'm safe. ;)

[trueMarketing]

I heard from a friend over at Yahoo that it had something to do with an algorithm close to the one Google has for page rank, but relating to indexed sites and new listings.
In other words something like:
P=pages, N=new pages, O=older pages, servers=# of total Google servers

(ID) Index Date = (P/N)+(P/O)/(#servers+(?)..... and so on...

This may be total crap, but maybe worth looking into since everyone is so into finding out when this happens. I will find out a little more about it as well as I can get info.

l8r

[bobmark]

Obviously I have no idea if this is true but it would make sense in terms of what seems to be this month's experience, i.e.:
1) the update is now approaching the upper limit (barring the time it took until the 6th of the next month) of the range of start dates which usually maxes at the 25th of the month;
2) myself and many on here have noticed very heavy crawling activity by Googlebot over the past week.
If, as your friend says, there is a threshold for update based on when a new index is a certain proportion greater (or, less likely) smaller than the existing one then you could see crawls increasing if, at the usual update period, the threshhold had not been reached and Google became concerned they had missed new activity.
Then again, it may simply be that the "minty fresh" changes have had unanticipated effects on either the "index in waiting" or their manpower and that is what is delaying implementation of this update.

[bether2]

My guess is that rolling out the update is similar to producing a monthly update to a software product. (A very intensive process!) There must be many different pieces of the "product" - the bot software, the algorithm software, the software that manages the data, and more.

For the algorithm piece, for example, I'm thinking that they must have certain new features that they want to add before doing an update. So the developers each work on their own piece. And then the testers (Quality Assurance or QA) test the software to see if it is producing the expected results. The QA team then lets the developers know what problems they see, the developers fix them, QA tests again, etc.

It is not unusual for a software product to take 50% (or more) longer than the projected time. In the case of Google, they don't have that kind of leeway.

If they're getting to the deadline and a certain feature can't be made to work as well as they'd like, it needs to be "backed out" of the code. By this time, the developer is probably is sleeping in his office at night - not to waste the time going home and back. And then, of course, the code needs to be tested again after the changes have been backed out.

And when software is ready, they apply it to the data and roll out the update.

I may be way off base, but this is my simplistic take on what the process might be.

Beth

[bobmark]

I think your probably about right, Beth. The only other thing I would think is that I am sure, like all large IT ops, they must have problems with "legacy sub-routines". You know...the kind of things that have been worked on by 50 programmers over 10 years and have become so complex that everyone is afraid to do a major rewrite even if the sub-routine causes problems that have to be corrected on an individual basis each time it's run.

[dvduval]

With over 2 billion websites in the index, I'll bet Google has quite a few computers running 24/7 all month. Not only does Google cache the pages, but it also establishes the link relationships between 2 billion pages. I'll bet that process takes about a month to complete. Maybe this time Google has increased the size of the index, because I've seen a lot more indexing of sites this month.

So my theory: Google doesn't just flip the switch. Instead, Google has to check each of the 2.4 billion pages againt the other 2.4 billion pages. If that's true, then we are dealing with an enormous integer, which I believe would be expressed as 2,400,000,000!
(note: the exclamation point. what you call that?)

[edited by: dvduval at 4:40 pm (utc) on Sep. 25, 2002]

[dvduval]

You call it a factorial. So can anyone figure the integer of the factorial 2,400,000,000!

Note: you'll need a REALLY FAST computer!

[bether2]

Yeah, you're probably right about that, Bobmark.

And when you start to think about the complex interactions that could occur between the "main" algorithm software, the bot software (including its own algorithm, I suppose), the data, and who knows how many other pieces, it sure makes you appreciate what goes into the whole process.

Not to say, that I, like everyone else, wouldn't like to see the update happen right NOW. : )

Beth

[bobmark]

I think one of the best things beginning webmasters can do is to install a "private" search engine on their site for internal site searching. Just by doing it you get a great feel for how se's work, how they see your site and what is involved in a re-index. It also has the benefit of showing you errors you've made (duplicate pages, etc.) before they get into Google.

[Helpmebe1]

I think some of you are hysterical.. I am not sure about the chanting thing.. would like to hear more about the yahoo source's info.. even though 2 or 3 days although anticipation kills us we have a good idea when its gonna happen.. we need to know crawl dates and approx dance dates... I think keeping an eye on when googleguy posts wouldnt hurt... he seems to post late local time so if theirs a 2am post..either hes stumbling in the door and just checking up on things or hes working late...

[musicales]

It's my time of month again. How many of you get PUT (pre-update tension) bad?

Google operates in a place beyond space and time, trying to understand a linear logic to the process is clearly missing the point. What we see every month (or so) are merely the off-shoots of a complex multi-dimensional space-time continuum. I think it probably has something to do with the emergence of new black holes in the universe.