Google Blogoscoped

Sunday, May 18, 2003

Google Fellow Urs Holzle

Googling Urs Hölzle — Google Vice President of Engineering since 1999, and since 2001, Google fellow — turned up an interesting interview in German, of which I will translate some parts:
Why doesn’t Google support a wildcard operator? Why isn’t it possible to use the query “cigar*” to find “cigar”, “cigarette”, or “cigarettes"?

Urs Hölzle:
Good question... mainly because this is relatively hard to implement and doesn’t always return the best results. Usually we try to keep Google as simple as possible, so that it automatically does the right thing. A better alternative would be e.g. to automatically include “cigarettes" for “cigarette”, or to treat “e-mail” and “email” the same. Our research group is currently investigating this. (...)
You are one of the founding fathers of dynamic “just-in-time” compilation. So regarding the basic dynamic of the Internet; isn’t it the case people these days put too much emphasis on the technical aspect? Shouldn’t usability be more important?

Urs Hölzle:
Absolutely. Usability is the key and often ignored. (...) It’s no coincidence usability guru Jakob Nielsen is part of our Technical Advisory Board. If a service is hard to use, even the best technical implementation is useless.
How will the Web technically evolve? Will it once reach science-fiction? Communication with a PC on a similar level — man to machine, so to speak?

Urs Hölzle:
That will take a while, if it ever comes to it. Of course the best search engine was the computer in Star Trek: just pose a question, no matter how complex, and the answer is served immediately. Unfortunately, we are still very far from really “understanding” text (...)
Small steps in this direction are possible (e.g. connecting the words “computer” and “PC”), and we are also working on this. More and more information will be available online, but the thinking will be done by humans for quite a while...”
Interview with Urs Hölzle, by Robert Herbig, January 25 2002

Google Cluster Architecture

Interesting in-depth article on the Google cluster architecture (written by people who work for Google):

“Amenable to extensive parallelization, Google’s Web Search application lets different queries run on different processors and, by partitioning the overall index, also lets a single query use multiple processors. To handle this workload, Google’s architecture features clusters of more than 15,000 commodity-class PCs with fault-tolerant software. This architecture achieves superior performance at a fraction of the cost of a system built from fewer, but more expensive, high-end servers.”
-- Luiz André Barroso, Jeffrey Dean, Urs Hölzle - Web Search for a Planet: The Google Cluster Architecture [PDF], April 2003 (IEEE Computer Society)

Question Phrases for SEO

If you run a website, you want people to find you. It’s relatively easy to know the right keywords. But do you have the best phrases? And what are the best phrases?

Let’s go through an example. Someone’s TV is broken. You got this extensive help site. Maybe you want to sell some books on how to repair the TV, or maybe you make money with advertisement. Or maybe you want to sell new TVs. Let’s analyze how a user could approach this search.

Note: Even though Google says it ignores the words “how”, “to”, “I”, “do” and so on in unquoted queries, it does in fact not. Just compare the search results of the (unquoted) queries “How to repair a TV” and “Repair TV” — they are different.

What happenedMy TV is broken
What do you wantI want a fixed TV
What do you want to doRepair my TV
What do you want to knowHow to repair a TV
What do you needInstructions to repair my TV
Who do you needSomeone who knows how to repair a TV

As you can see, these are totally different. I entered some phrases into Google and here you can see how many result pages I got:

Query Results
PhrasePages returned for quoted phraseFirst result for the unquoted phrase helps
Repair TV 29,800No
Fix TV 4,220No
Broken TV 2,730No
Fixing the TV 285No
The TV is broken 266No
My TV is broken 151No
TV repairing 144No
Fixing a TV 58No
Repairing the TV 47No
Repairing a TV 40No
How to repair a TV 11Yes
How to repair the TV 3Yes
How do I repair my TV 0Yes
How you can repair the TV 0Yes
How can I repair a TV 0Yes
How can I repair the TV 0Yes

Look at the sentences with the least result pages — in other words, a nice phrase that would make sense to target if it’s actually used. They are returning the best pages. Can you see what the sentences have in common? They are all in question form. People making websites often think about what they got — forgetting what the people trying to find them want.

I’d never suggest to a webmaster doing Search Engine Optimization solely for the Search Engines. (When actually, it should be optimizing accessibility for the end user.) However if you keep in mind to think from the visitor’s perspective, you might find more varied phrasing, you can alternate titles, and thereby enrich your text. You can also enhance the site by adding a FAQ:

People are getting used to asking questions and putting questions directly into search queries because of the “Ask Jeeves” push. However, we have noticed that referrals coming through from Google where questions are asked in the query has increased significantly. So we tried the FAQ concept. (...)

The most common questions people ask start with:

How do I . . .

What is . . .”
Microdoc on FAQs, May 2nd 2003 (I-Help-You Google Forum)

For our specific example, the page could go like this:

“How to repair a TV

When your TV is broken and you ask yourself, “How do I repair the TV”, read on here.
Repairing a TV consists of several basic steps, outlined in the instructions to repair a TV below."

Quite obviously a highly constructed example (and not a very nice looking one at that). But an introductory sentence describing what you want to say on the page can’t hurt. And it makes sense to not repeat the same title over and over again. Usually, the text will be much longer, so you don’t have to artificially boost phrases like done here.

Google Answers Researcher Interview: Tlspiegel-ga

The interview today: Tlspiegel, a member of the Google Answers Research Team since the week the project launched.


What’s your first name?
Toby Lee (and that is my entire first name, my parents were to poor to give me a middle name!)

Where are you from?
Born and raised in Pittsburgh, PA — lived all around the Midwest and now spending the balance of my wonderful life on the Heat Island known as Phoenix, Arizona.

What’s your profession?
I’m conditionally retired. I’ve worked at many jobs, but never had a career other than raising a family and animals.

How old are you?

Do you have a homepage?

Tlspiegel-ga on Google Answers

What was your first answer for Google Answers (please include date)? What was your most recent answer?

What kind of questions do you like to answer most?
Where can? How does? Who said? What is? If a question interests me I’ll lock it.

What were some of the most interesting discoveries you made during your research activity (please include the questions)?

Hypothetical Musings by Tlspiegel-ga

You have one minute to convince a potential customer who doesn’t know about Google Answers to start using it. What do you say?
At least once a day I will improvise on the following:

You can post a question to Google Answers ranging from $2 to $200. Within approximately 24 hours, one of many researchers will search for the answer and post it to Google Answers.
As researchers, we have style guidelines and Terms of Service we must follow. You can read all about it at this url:
We’ll use a variety of resources — some I’ve used would be searching online, library, personal experience, books, and even make phone calls to verify facts.

If an omniscient deity would be a Google Answers Researcher, which question would you ask?
“Please explain the very beginnings of creation.”

Which famous person, dead or alive, would make a great addition to the Researcher team?
Sigmund Freud

Time magazine features Google Answers; what can be seen on the cover?
The word “BETA” with a giant Red X through the letters.

Tlspiegel-ga’s Favorites

What are your favorite research tools, on- and offline?
I don’t know if I could manage without: MyIE2, Clouseau’s bar, Metapad and a Spellcheck Utility.

What are some remarkable less known websites?
I don’t know how remarkable or how unknown they are but here goes:

Tlspiegel-ga’s Spare-time

Got any weird hobbies?
No weird hobbies, but I have an interest in the paranormal (I really do miss Art Bell). I’m a Guide at a Forum which keeps me occupied. I take care of my pool. And, I am fascinated with a project my brother has been working on for about 18 years.

What are some of your favorite books, movies, and music albums?
I have so many favorite books, but one I’d want to have with me if I was stranded on an island:

Favorite movies: “Fargo”, “The Bird Cage”, “Midnight Express”, and an HBO special “Robin Williams Live on Broadway”.

My music would have to include:

Final words by Tlspiegel-ga

What would be the title of your autobiography?
Ha! I Told You... I’d Survive

Anything else you might want to say?
Please inscribe the words: ’Google Answers Researcher’ on my tombstone, and if there’s limited space just put GAR!

The Google Society

Today in the New York Times, Geoffrey Nunberg talks about the Search Engine Society: As Google Goes, So Goes the Nation. Society needs religion, and religion needs commandments. Here goes:

The Ten Commandments of Google

  1. Thou shalt have no other search engine before me.
  2. Thou shalt not enter more than ten keywords.
  3. Thou shalt not take the Google trademark in vain.
  4. Remember the Google URL, to keep it holy.
  5. Honour thy Brin and Page.
  6. Thou shalt not have low pagerank.
  7. Thou shalt not commit linkfarm.
  8. Thou shalt not write the OR operator in lowercase.
  9. Thou shalt not quote from the Google cache.
  10. Thou shalt not press the I’m Feeling Lucky button unless thou feelst lucky.


Blog  |  Forum     more >> Archive | Feed | Google's blogs | About


This site unofficially covers Google™ and more with some rights reserved. Join our forum!