Forum Moderators: open
I hardly ever visit here anymore, mainly because all the Great Google Theories of Yore seem to have been shot to pieces, and all the hocus-pocus promulgated by the Almighty Google Gurus (at whose very feet I once worshipped) seem to have come to naught. Or at least the don't carry the same authority they once did.
Oh sure, there are still some great tips to be picked up in this fabulous forum, but I'm finding more and more of them outside the Google News section.
I do miss the 'old days" though. I miss the certainty of the "update" and sitting here - fully stocked with booze and ciggies - watching and waiting as the update spread across the datacenteres... and refreshing this site every 10 seconds, just in case I missed more that 20 new postings.
But maybe I'm just getting old. :)
Not hardly wanting! LOL...
According to the active list we had close to 40 active threads in this forum alone today with loads of posts each, and Friday isn't generally the busiest day. It's a LOT to keep up with.
>>finding more and more of them outside the Google News section
Instead of being all together like Google Stew, some are being moved to other appropriate forums if they're applicable, which some are. Oh, and some gets posted to the other Google forums - to the Toolbar, Business/IPO and AdWords forums.
I kind of miss the excitement of the update threads, too - when there really was an update. Those were exciting and fun, kind of like a monthly celebration. I'd generally put coffee up and stay up all night watching along with everyone and checking the SERPS. I do miss that. But I don't miss many posts a day asking if it's starting or when it will be - or is it over yet.
There's plenty to read now, and it's much calmer so it's easier to really dig in and think things out - and actually get other things done, too.
I can't believe a search engine can be so considerate and actually factor in people's well being too.
With some of the rules perhaps having changed or evolving, makes it all the more necessary for a forum to discuss new tactics. Plus the clever people here (I am not one of them) who worked out how to do well before, will do so again.
I do miss the 'old days" though. I miss the certainty of the "update" and sitting here - fully stocked with booze and ciggies - watching and waiting as the update spread across the datacenteres... and refreshing this site every 10 seconds, just in case I missed more that 20 new postings.
Even though I only witnessed a handful of pre-Dominic updates, I do miss the big show.
It reminds of the old Jerry Grateful Dead shows. I didn't latch onto them until their last few years of touring, but I was fortunate to have been there for a few rounds.
I posted a new topic that was very relevent to an event on Friday where a Google employee was speaking. Never heard as to why it wasn't approved. See, that had very little to do with this thread, but that's what we're left with.
There's also the element of the unknown that has probably reduced the usefulness of Google News. Since nobody really knows what's going on right now, there's not always much purpose in discussing it.
That was your 55th post and you never visit here anymore?!
Was there anything interesting said at that event?
Actually, that was my question. Here's the only summary I've found so far:
Nelson Minar of Google talked about a lot of aspects of Google, but focused heavily on their advertising. To their mind, Google is not just about searching the web, it's about presenting information that's useful to people. To that end, they consider successful advertisements to be another form of search problem: you have to consider relevance and user experience. That's why Google separates ads from search content (you can't buy ranking in the search results, period), uses complex algorithms to determine which ads are appropriate, and (most importantly) presents the ads in user- and bandwidth-friendly text only. It's surprising how well received Google's ads are in the "everything should be" crowd. It's thoroughly impressive how much work happens in those 0.2 seconds of search time!
You can search for yourself [google.com], too. I'm sure more will come up as google finds it. Apparently there was a very interesting Q&A after the talk.
EDIT, here's some more:
I recognize this guy, he was in a Fast Company article a while ago. See it here [fastcompany.com].Main Index - indexed once a month.
Fresh Index - indexed once a day.
News Index - indexed every few minutes.
The above indexes are used to return resulsts produced by the Google search algorithms - which are way too complex to explain (and they are secret).
Google News - searches over 4500 sources. The front pages ranks stories on freshness.
AdWords - Ranking = Max CPC x CTR. The Click Through Rate does matter. If you pay more but don't get any click throughs, you will end up lower on the list.
AdSense - keywords, frequency of keywords, font sizes and word placement, anchor text - all determine what ads are shown.
The Google Datacenter - 80 node clusters, everything redundant.
google: relevance and results.google mission: organize the world's information to make it universally accessbile and useful.
our goal is for you to feel the info you find is relevance. the work we do is driven by this mission.google philosophy: work on things that matter. affect everyone in the world. solve problems with algorithms if at all possible. hire bright people and give them lots of freedom. don't be afraid to try new things.
"no HAND work" - can't deal with that. let's solve with engineering tools.
empowering the folks who work here. great example is google news. started with just one person.
when companies get older they often get into a niche they can't break out of - we don't want to be like that.history:
google in the garage - started in a menlo park garage
simple design. theory #1: better for users. theory #2: we weren't that good at html
we believe simplicity of ui makes us very effective.we're there's electrivity there's google.
massive localization - 88 languages!
burundi; american samoa; gambia jersey....anatomy of a search result page:
news
search
advertising
bragging (how long it took us to find your info)google index: how does it work
three component
news - rebuilding each five mintues; special crawl just for news results
fresh - once a day; sites that change a lot; sites that people care about; some porportion of results to reflect recent
main -query comes into the google web server and we have 1/5 of a second to figure out what to return.
(image of this process to be posted)
index server gets the query first
index servers are replicated and distributed
pieces of the index scattered acrossdocuments servers
these actually have the copy from the web
again replicated and distributed
doc servers do their job and return the datamisc servers
ad server - query ads; relevant advertising
spell checker server- "did you mean....."
news serverNew Technology
crawl 4,5000+ news sources
clustering tech identifies same story in multiple locations
news front page synthesized entirely algorithmically
localizing news to different countriesgoogle advertising:
reason no banner ads: too slow to load
ads makes sense when you do a commerce related search
you get the choice - the commercial stuff and the research/consumer stuff
sometimes search results are not that useful as related to the adsrelevancy, pricing, and position
ranking = max.cpc x ctr
an ads position on the page is determined by performance and price:
we reward advertisers with good ads
we rank in an auction model
the more they are willing to pay often the better the result: cost per click
how often people click on it: click through ranking which is the users measure of relevance
the combination of the two let you know what is relevant both in the mind of the advertiser and the user
poorly perfoming keywords automatically disabled(aside: calling all math majors! you would love to work at google)
if an ad is below a .05 click through rate - we just don't show it.
advertisers only pay when people actually click on the ad - not just shown for view
we have created a market for advertising that helps the users find relevant data and advertisers showcase their productsChallenges: billing and syndication
ads much be servved quickly and reliably
- syndication partners, googl users don't tolerate downtimereal-time auction for each ad shown
log data is precious
- each click is only worth pennies, micro-payment accounting(random aside: how much do i love frank?!)
google ad sense
we understand a page based on such factors as:
keywords
frequency of workds
font sizes and placement of words
anchor text
linguistics processingwe put all these factors together and some magic software boils it all down to this.... okay this page is about x so the add we serve up needs to be about y. usnews.com is an example.
think about adsense targeting.
word java:
is it drink or programming language or the island?
find a second word: cup
java + cup equals more than likely the beverage.
google datacenters
pics to come of these:
goodgle.stanford.edu
earliest disk cases - legos
google servers 1999
google servers 2000
google servers todayhardware architechture slide
key problem is fault tolerance
pcs are unreliable, especially if you have thousands
but they are cheap and fast - linux the key
strategy: exploit procession power of off the shelf pc hardware, make it reliable in softwarewe use software as the fault tolerant.
conclusion:
losts of interesting and hard problems
focus on users, give relevant results
don't be afraid to try new thingsb.
I'm going to link [flashgoirl.blogspot.com] to this and hope that's OK, since it has more real info.
Only if you feel you should and have one available (a hole that is) ;-)