Wednesday, October 20, 2004

How to calculate ROI for intranet searchs

This combined hardware and software rack box is your next company' search engine, your Google in a box": The search engine inside is similar to Google's search site, aimed for the enterprise market. It runs on Linux operating system inside Intel box. (Apparently Google rolled it out in Europe last week

Designed for Intranet users, it fetch results in a format that is very similar to Google's SERP, including small excerpts and links to cached versions.

Want to give a try but can't find a way to convince your boss?

I've been talking about findability [French] on my French blog recently.

"[Findability is] the quality of being locatable or navigable. At the item level, we can evaluate to what degree a particular object is easy to discover or locate. At the system level, we can analyze how well a physical or digital environment supports navigation and retrieval."Morville

A lot a time is spent searchin' the web. A good corporate search engine should help prevent loosing too much time retrieving informations, especialy if it is for inside-the-firewall documents.

But how do you calculate ROI to be eable to buy a search engine?

Quite easy :

1. Let's say there is three categories of users. Heavy users; moderate users; light users.

2. Let's say category represents 20%, 30 % and 50% of the company. (these numbers are quite realist)

3. Let's put how long they look for information : 100 ; 50 ; 25 (let's say it is per year for each category of users).(average milage may vary.)

4. That's make for a 100 employees company, roughly for each category this much time spent (number of hour spent times number of employees): 2000 hrs; 1500 hrs; 1250 hrs. That's the number of hours per year spent looking for information in your 100-employee company.

To simplify the math, let's say all employees get a 50,000$ annual salary, that is a roughly 25$ per hour.

How much does it cost to look for information? That is 50,000$ for the first category(20employees x 100hrs x 25$/hr); 37,500$ for the second category; 31,250$ for the third category.

That's make 118,750$ per year

Let's put here a percentage of time saved with a good corporate search engine : 50% (that'a a guess). That's is a 59,375$ savings. Let's round it to 60K$ per year

You're now allowed to calculate the pay back period. My information told me that the low-end Google box start around 40K$ for a 2 year license.

That's rougly 3 months pay back period per two year. Now you can play with the numbers: time saving is half lower? double the pay back period. Employees receive 100K$/year?, cut in half the pay back period...

For knowlegde workers this is a good ROI proof to show to your boss. If Google is your primary internet tool, you might want it for your intranet too.

( I'm not affiliated with Google or any Search engine company )

Thursday, October 14, 2004

Wikalong, collective web page annotation

Wikalong, it is a simple, clever, wiki in your browser sidebar for collective annotations of Web pages.

Wikalong is an FireFoxExtension (Firefox is a hype navigator right now; compatible Mac, Windows and linux).

Wikalong appears in your sidebar and indexes the web pages you surf: ywant to make an annotation? open Wikalong, write your note and save ! That's it, that's all!

Web page annotation is not a new idea, but the use of the wiki for this goal is quite clever. Wikalong, allows you to see the notes left by the others as well.

The practical application?

If you build a Web site or you visit pages. You want to share corrections or comments! Instead of infinite email exchange (which you loose track of pretty soon) or instant messaging (which of course you'll forget) or posting on an external wiki (which URL you can,t remember). Wikialong do associate a wiki page with the Web page. Web page validation is maid easy!

The teaching application?

The professor leaves annotations like : word definition, related URL, contextual explanations (history, policy, technical), exams questions, etc.
The pupil leaves annotations like: vocabulary questions or meaning ambiguity questions, related URL, quotation excerpts, etc.

The commercial application?

I see a bright future for the wiki here: as wikalong web server is public, you will want to pay to have your private workspace. The wikis farms should offer this extension, this may bootstart wikis hosting industry!

Installation:

No inscription, not software installation.
1.install Wikialong extension (http://wikalong.phunnel.org/wikalong-current.xpi (for Firefox)
2. Restart Firefox
3. Go to menu::VIEW::Sidebar::+Wikialong
4. Click on my permalink and you will see the comments associated with my web page.

Wednesday, October 13, 2004

Future of Real Estate

RedFin this will change forever the way you look to buy/sell a house. An internet real estate tool, with flash interface.

With satellite photos, overlayed with city and street information, then overlayed with information on houses for sale : here's your RedFin tool to buy a house. Forget all other DB-driven tools, this really rocks. (Seattle area data only right now) .











--
Link: Redfin : http://www.redfin.com/
Via : Karl Nelson : http://www.karlnelson.net/weblog/000853.html
Feench Version: http://zeroseconde.blogspot.com/2004/10/redfin-limmobilier-et-le-virtuel.html

Tuesday, October 12, 2004

Clickwrap nightmare

"I have a recurring nightmare. Microsoft CEO Steve Ballmer shows up on my doorstep demanding my left kidney, claiming that I agreed to this in some "clickwrap" contract."

Mark D. Rasch in Wired 12.10

Friday, October 08, 2004

Anti-Spammer Honey-pot Email

Wondering about how to build a kind of email Immune System?

Set up honey-pot fake email addresses on your site which nobody is meant to mail to. Any email which arrives at both this honey-pot and your real email address, is subsequently deleted from all accounts that this server handles.

Read on on nooranch.com wiki

Monday, October 04, 2004

Finding data instead of knowledge

French language doesn't make any difference between "information retrieval" and "search" : "Rechercher" means "to find information", with no clear cut betweens the means and the goal.

French dictionary does specify a difference though between computer science's definition ("function that retreive specific datas") and information science's definition ("Work tool consultation of archives to seek information"). At least this is a clearer cut between means (computer world) and goal (communication goal. And this may show why, by the way, programmer and communication teams don't get well together)...

Information Retrieval uses search engine (like Google) with the aim to fetch relevant information. Wait, shouldn't we say data instead? And what about knowledge finding?

KM (knowledge management - but I prefer Knowledge work management) got one objective : finding /producing new knowledge.
"Knowledge Management is not about acquiring more information. It is about expanding the ability to get results - even with the same information".
C. Jackson Grayson, chairman and founder of the American Productivity & Quality Center (APQC)
(via Florian Heidecke)

Meaning is created through a transformation processs :
  • Data : stocked on material
  • Information : Formal data
  • knownledge : formal information / ideas
To google isn't a neutral verb : the SERPs is highly mathematical but show nonetheless a world representation that isn't neutral. Classification, ontology, knowledge organization do come from a mind that discreminate data to make information relevant to him/herself. Nothing tells you that mind got your very own world's point of view. SERPs do change the way you see what is available, thus what the world (or your corporate memory) has for you.

And there come Serendipity and google pagerank as the new paradigm to fight information entropy and findability, but this is another story.

(French version of this post)

Saturday, October 02, 2004

Bush-Kerry debate linguistic stats

Interesting debate linguistics stats from Lucas Gonze's friend Anatole :

Bush's use of contractions: 4.4% of words.
Kerry's use of contractions: 3.6% of words.

Bush's average sentence length: 12.681 words.
Kerry's average sentence length: 14.896 words.

Bush's average sentence fragment length: 7.251 words.
Kerry's average sentence fragment length: 7.677 words.

Bush's spoken words: 6332 words.
Kerry's spoken words: 6696 words.

Bush's vocabulary: 1032 unique words (de-stemmed).
Kerry's vocabulary: 1076 unique words (de-stemmed).


More stats on his post : http://gonze.com/weblog/story/10-1-4#10-1-4