Re: Sending Out An SOS

[Stuart Hofmann]

Eugene Miya wrote:

>
> >Not my fault that Google doesn't know about 12th century Knife technology....
>
> The trick is knowing how to use Google to find information, but it is
> not perfect.
>
> The issue is not necessarily to perform a Web search.
> The way to think of google is as an index; it has no intelligence in
> the human sense yet (and does have some in the military sense).
> The problem is that the Web is merely a portion of the Internet;
> Usenet is a different portion of the Internet which google also has access to.
> Google, as a service does have other information sources.
>
> Don't mistake the Web for google or vice versa.

"Kukkerie" was obviously ~wrong~. Approaching Google head-on was sure to
disappoint. The way to find the info was to search 'nepal' 'knife'.
-S-

[Gary S.]

On Wed, 11 May 2005 01:13:40 GMT, Stuart Hofmann >
wrote:

>Eugene Miya wrote:
>
>> >Not my fault that Google doesn't know about 12th century Knife technology....
>>
>> The trick is knowing how to use Google to find information, but it is
>> not perfect.
>>
>> The issue is not necessarily to perform a Web search.
>> The way to think of google is as an index; it has no intelligence in
>> the human sense yet (and does have some in the military sense).
>> The problem is that the Web is merely a portion of the Internet;
>> Usenet is a different portion of the Internet which google also has access to.
>> Google, as a service does have other information sources.
>>
>> Don't mistake the Web for google or vice versa.
>
>"Kukkerie" was obviously ~wrong~. Approaching Google head-on was sure to
>disappoint. The way to find the info was to search 'nepal' 'knife'.
> -S-
>
A wise person knows how to restate a search like above, to get good
results.

URLs and Google will force people to spell correctly, although a
number of sites pop up with common misspellings as well as the correct
one (good design).

The mistake that many make is in assuming that the totality of human
knowledge is accessible on the Web. Still a large place for the
traditional info sources.

For example: US patents from 1975 and later are online, and fully text
searchable in a number of ways. US patents from before 1975, all the
way back to 1792, IIRC, are only indexed by patent number, and only on
the Web in image form. You still may need to use the printed indices
at the 10 public libraries around the country which have older US
patents on microfilm. Patent searches on older ones are significantly
more difficult.

Many databases and archives are subscription only, and do not show in
a Google search. Older magazine and newspaper articles are not going
to be put online, although a selection of newer ones are tehre, and
some restricted archives going back a short time. Hundreds of millions
of printed books are never going to be put on line either.

It will be a long time, if ever, before one can do a completely
thorough, Net-only search.

Happy trails,
Gary (net.yogi.bear)
--
At the 51st percentile of ursine intelligence

Gary D. Schwartz, Needham, MA, USA
Please reply to: garyDOTschwartzATpoboxDOTcom

[Eugene Miya]

>>> >Not my fault that Google doesn't know about 12th century Knife technology....

Eugene Miya wrote:
>>> The trick is knowing how to use Google to find information, but it is
>>> not perfect.
....
>>> Don't mistake the Web for google or vice versa.

On Wed, 11 May 2005 01:13:40 GMT, Stuart Hofmann >
wrote:
>>"Kukkerie" was obviously ~wrong~. Approaching Google head-on was sure to
>>disappoint. The way to find the info was to search 'nepal' 'knife'.
>> -S-

This is a decades old semantic information retrieval query.
A lot of US money was spent and is still spent to second guess people
making queries like this. And maybe one day, irrespective of the
approximation, the querier will get what they seek. BUT that will be at
certain costs.

In article >,
Gary S. <Idontwantspam@net> wrote:
>A wise person knows how to restate a search like above, to get good
>results.

Librarians think that. The reality is that this technology still has
development issues. One thing that you guys who never have a chance to
visit the Googleplex campus is that each of the visitor lobbies project
a select sample of current web queries. Over time they had to add
non-7-bit-ASCII character sets. They remove certain explicit queries
of overwrought sexual content, and a certain fraction of simple queries
get ignored and "interesting" queries get selected. Groups doesn't get
this treatment. It is somewhat entertaining to watch this sampling
of queries go by. Google retains all queries. They retain as many of
these posts, they reserve the option to scan your gmail accounts.

Right now they are good guys, and changes and biases are very subtle.

>URLs and Google will force people to spell correctly, although a
>number of sites pop up with common misspellings as well as the correct
>one (good design).

URLs will, Google won't. They are very aware of this (this is referred
as the B. Spears problem, you can find their web data examples).
Google still doesn't do what is called "stemming." A plural, a prefix,
a suffix will still make differences. So I have no idea what you mean
by good design here Gary. My old Divison Chief composed the Spears
table of misspellings (over 100). I have a similar, pre-Web table from
the Library of Congress "misspellings" of Gaddafi (1 of about 60).

I was a far worse speller before computer text editors and checkers.
But I won't pay my copy editor for posts. She works for the print journal.

Never forget that spelling is a convention.

>The mistake that many make is in assuming that the totality of human
>knowledge is accessible on the Web. Still a large place for the
>traditional info sources.

The mistake is in presuming that humanity knows more of the universe than
it does. None of this info is in traditional sources.


>For example: US patents from 1975 and later are online, and fully text
>searchable in a number of ways.

You still can't search on photos, images, music other than textually,
mathematics, chemistry, etc. Foreign language support is dependent
on reader understanding and knowledge. In this way the English
will continue to manipulate US Americans.


>Many databases and archives are subscription only, and do not show in
>a Google search. Older magazine and newspaper articles are not going
>to be put online, although a selection of newer ones are tehre, and
>some restricted archives going back a short time. Hundreds of millions
>of printed books are never going to be put on line either.

Well, we are working on the latter.

Subscription bases like Lexus/Nexus, Dialog's INSPEC, etc. are
being given a run for their money. While it is hard for humans (John
Henry) to compete with Web crawlers (machines),
the slow addition (and correction) of web services improve to low-level
"good enough" queries (a long term problem). This subtle difference
is why Google page rank is better than prior search engines. It's web
changes, not search algorithm.

It's not merely google. Amazon, Yahoo!, Barnes and Noble, etc.
all contribute to this.


Oh, sure, for the forseeable future, it will appear like Achilles racing
the tortoise. Or say Kasparov vs. IBM Deep Blue playing chess.
Oh, but that machine won. But I can't credit IBM for having AI.
The important thing isn't to get into useless linguistic pardoxes like
Achilles.

>It will be a long time, if ever, before one can do a completely
>thorough, Net-only search.

Depends on the query, the person asking, and the need.
And whether the query is good enough.
For example, since you and I are posting in r.b., your failure to edit
panel 7 of FAQs merely contributes to delaying time.
And that editing is instructive to search problem.

Google is hiring.

--