Matt Cutts: Panda data is not updated daily right now

[suzukik]

These are conversations on Twitter seemingly related to Panda Update.

@mattcutts I can word the question better in >140 chars :) so far I don't know of any site that's recovered traffic. Would be good to know

via @tomcritchlow [twitter.com]

@tomcritchlow answered that for NPR, but they didn't use it. Answer not as amenable to 140 chars.

via @mattcutts [twitter.com]

@tomcritchlow short version is that it's not data that's updated daily right now. More like when we re-run the algorithms to regen the data.

via @mattcutts [twitter.com]

Dose this mean Panda is updated at relatively long interval?

[whatson]

I agree

[tangor]

Maybe they can't do it daily or on the fly

I think folks are over-thinking this. Google (after Caffeine) has had the ability to crunch big time (after all, they told us so). What the PhDs at Google have done is introduce so many different indicators into their algo that even the mighty Caffeine can't keep up. After all, how many different ways can you say (find, see, taste) "crimson", which we all know is "red". Unless you're determined to define ochre, rust, blood, blush... Soviet, China, clay, Zero, etc. to extend your "red" radar out the wahzoo.

Google is broke. They have accumulated so much they can't figure out what to do with it (after abandoning the original intent of INDEXING the Web... i.e. words and value of which in relation to near content.

Google is broken at the moment. It has been "fixed" beyond repair in the search of fixing a thing that is functional until it is unusable.

Much as I love Bing these days, I see they are following, if NOT LEADING in some aspects. After all MS has more dollars and PhDs to throw at a problem that really does not need to be solved.

I bring no solutions to what we seek, only a rant of the obvious. Forgive me. I know what I seek... and I'm not finding it these days...

[walkman]

tangor, you might be onto something. That's why they are choosing a handful of 'safe' sites to get virtually all the traffic for each niche.

[tedster]

There are other components to Google Search that are not updated daily - the n-grams used in phrase-based indexing, for example, or the automated taxonomy generations. I'd predict that Panda will, eventually, be integrated into a more dynamic update process - but for now, we have another sign of just how complex this algo component really is.

[walkman]

Tedster - yes or no :)- in your opinion, has Google updated the quality score or whatever is called for the sites hit on 2/24?

The problem is that the algo is brand new, and it is very irresponsible if Google heavily penalized sites based on it, without taking in consideration any changes for over 3 months.

[deadsea]

I'd be very surprised if Google didn't run their calculations for it at least monthly. I suspect that the continued drop to many sites that have made positive content changes is due to the devaluing of inbound links from other sites that have been pandalized.

[tedster]

in your opinion, has Google updated the quality score or whatever is called for the sites hit on 2/24?

Yes. We know they changed the algorithm twice - that certainly means it was run again for 2.0 and 2.1. And from other ranking changes that some impacted sites have seen, I think some version of Panda scoring was recalculated at other times when the algo itself remained unchanged. How often is hard to say right now. Google is being less than transparent (as we've commented on) and certainly other factors in the algo are also shifting.

If I were pushed to guess, I'd say every 2-3 weeks. But I'd also say that there's something provisional about any improved score - it gets tested and rankings get changed step-wise. That's guesswork from watching rankings.

[walkman]

Interesting. Maybe it's wishful thinking on my part and many others ;). I find it interesting that the question started with "so far I don't know of any site that's recovered traffic..." but then doesn't mean it didn't happen and that was May 7th. I rank for the same usual page specific keywords I have since Panda, "widget domain name" type and now there's nothing special about those pages at all, compared to my other pages. Before they *may* have had had a bit more content and I have many more less competitive ones that in theory should rank easier. The pages I get traffic from are very predictable and so are they keywords despite my content chage, but some days I may get more, some days less.

The only way I see Google doing this for three+ months is to teach 'bad sites' a lesson and since the worst of the worst we bundled in 2/24 Panda they let everyone wait /suffer. But Goog must have added some criteria based on sites caught by mistake and that posted on Google threads, so that had to be rectified. Run the entire algo or part of it? What if a notorious content farm escapes and Google is embarrassed like they were with eHow? We all were "eHow escaped and my wonderful site was..."

I guess we'll find out...one day.

[whatson]

It is possible that they are making people #*$! themselves, just for the sake of extra "motivation" to make their sites that more squeaky clean, and remove absolutely anything that might potential harm them, and do whatever they can to improve their visitor satisfaction. Giving that omniscient presence to scare us, when in fact they don't know quite as much as we hypothesize.

[HoosierBuff]

I do wonder if the sites hit weren't hit with a minimum of a 90 day penalty - 90 days seems to be a duration that Google likes.

So, the scenario would be:

Slapped by Panda on 2/24,
Site owner makes changes (3/15)
Google reruns algorithm (april 8)
90 day anniversary of panda 5/24/11
Site owner sees relief 5/25?

[walkman]

@HoosierBuff
I will buy a nice cold (case of) beer if that's case. That's my theory too...a fixed time penalty, but then I have covered every theory here since Panda bit mauled me. Goog would probably run the algo again, not rely on April score.

Giving that omniscient presence to scare us, when in fact they don't know quite as much as we hypothesize.

Well we know how they can tell everything ;) One day #1, next month #61 and then back up to #2 without changing a word (exaggerated a bit but ...)