Home Messages Index
[Date Prev][Date Next][Thread Prev][Thread Next]
Author IndexDate IndexThread Index

Re: Google Trends

__/ [ Doc O'Leary ] on Friday 12 May 2006 20:54 \__

> In article <2763555.9OSNo8546l@xxxxxxxxxxxxxxx>,
>  Roy Schestowitz <newsgroups@xxxxxxxxxxxxxxx> wrote:
> 
>> To select keywords you will be better off using:
>> 
>> http://www.google.com/webhp?complete=1&hl=en
> 
> You clearly haven't done any sampling/comparison between the two.  In
> truth, what Suggest offers doesn't make any real sense.  For example, if
> you start with "mac " you get a number of Macintosh related completions,
> but if you chart them on Trends you see:
> 
> http://www.google.com/trends?q=mac+mini%2C+mac+games%2C+mac+tools%2C+mac+
> os+x
> 
> Notice that I added "mac os x" myself, and it completely dwarfs the
> things Suggest was feeding up.  I have no idea what algorithm Suggest
> uses, but it clearly isn't based directly on what people are searching
> for. ... [par broken]


No,  it  is based on the number of results. Judging by a rather  extensive
use of Google Suggest, they seem to have not updated the numbers for about
a  year.  These were snatched off-line at some stage and they  potentially
motivated  this  'sequel' from "Google Labs" (whetever this  truly  refers
to),  which they called Google Trends. Many predicted such a service -- an
insight  into  Google  logs with a national  information  breakdown  (will
become controversial IMHO).

Maybe the features in Google Suggest just merely 'fell asleep', going into
hibernation   and   serving   as  a  nice  prototype   for   Google   News
auto-compeltion.  GoogleBar  incorporates  this as an  experimental  (thus
fragile) feature and maybe Google Toolbat likewise. I can recall something
in the press of the Google Blog about it...

[par resumed]

> ... Suggest might help you find a keyword or two you might not thought
> of (if you're willing to run through the alphabet!), but when it comes
> to actually paying for one, you'd be a fool not to plug them into Trends
> first to compare the results.


It's  something  I mentioned in the search engine groups yesterday.  These
can  easily be tweaked, SPAM'med, manipulated, subverted, SEO'd... call it
what  you  will, but it gives people a grip on statistics. And what  about
tools  that  automatically  query  the  datacentres?  Like  Perl  scripts,
front-ends  for SEO purposes and the like. This raises more questions than
it  answers.  Other points to ponder: are figures and chart normalised  by
overall use of Google as a search engine? City's population size? National
figures? Are all (wo)men count equally?

Best wishes,

Roy

-- 
Roy S. Schestowitz      | "Computers are useless. They only solve problems"
http://Schestowitz.com  |  Open Prospects   ¦     PGP-Key: 0x74572E8E
  4:55pm  up 15 days 23:52,  12 users,  load average: 2.02, 1.92, 1.67
      http://iuron.com - knowledge engine, not a search engine

[Date Prev][Date Next][Thread Prev][Thread Next]
Author IndexDate IndexThread Index