Armstrong Consulting
1200 Dale Avenue #100
Mountain View, CA 94040


Date: Wed, 03 Oct 2001 15:35:24 -0700


From:   Eric Armstrong
eric.armstrong@sun.com
Reply-To: unrev-II@yahoogroups.com

To:     unrev2
unrev-II@yahoogroups.com

Subject:   Recommendations??


Jack P:

You've posted some 400 messages to the list (or maybe it was 500 -- I started losing track after 300 or so), most of which point to some impressive, cool, or potentially useful technology.

John D:

You've posted in the neighborhood of 200 messages, most all of which point to some powerful, cool, potentially useful technology.

These pointers would be very useful, if I had anything like the kind of time it takes to track down some 700 relative technologies, understand what they are about, and figure out how they can be applied.

However, I do not have that time. And as great as it is that you keep finding new, interesting, useful, and cool technologies, I find myself realizing that I am never going to be able to know how the latest revelation compares with, or may possibly interact with, any of the other 700 recommendations.

The "information explosion" exhibited by these pointers alone illustrates some of the *vital* requirements for a useful collaboration tool:

  1. Categories

    When recommendation "X" comes in, it needs to come in with a category (or multiple categories) or, better, categories need to be retroactively applied, so I can tell which recommendations achieve similar goals, or perform similar functions.

  2. Ratings

    There is no way on God's green earth I am going to investigate 700 recommendations, until and unless that is my paid job function (at which point I will be more than happy to undertake the task).

    Until that I occurs, I *must* have ratings for these things, so I can idendify "best of breed" in each category.

  3. Combinations

    If someone can say, "we can combine technology X with technology Y to do Z". That new combination can then be categorized and rated, so it can be compared with combinations X and M, or combination M and N and P.

At this point, I find myself in the exact same position as the CIA. Someone will always be able to say, in retrospect, "see, I told you it could be done using X", for any "it" and an "X", where "X" is one of more than 1,000 alternatives that are buried in the list, once everyone's contributions are taken into account.

However, the current system will only allow that recognition to be achieve retroactively. When one person with a limited number of technologies at their disposal figures out how to make something work (because they aren't spending their life evaluating alternatives), then it will be clear that "we had the information" in our possession.

However, our ability to proactively identify that solution by examination of the combinatorial explosion of possibilities before us is negligible, at best.

A system that allows for categorizing, rating, and creating new combinations will allow that proactive identification of solutions, because any one person can contribute a small quantum knowledge (consisting of a combination or a rating), and that quantum can be compared with other relevant quanta (via categorization, which juxtapose related bits of information).

Without such a system, I find myself in a hopeless quagmire. There are too many options to consider, so "paralysis by analysis" becomes a real threat, were I ever to feel optimistic enough to attempt a foray. Given that any one combination is likely to prove untenable, the only way to feel optimistic enough to make an attempt is to know that, even though my approach will most likely fail, the result will be knowledge added to the system that help others steer clear, and the expectation that with all of us evaluating one combination or another, *some* combination may very well succeed.

But, absent the ability to share my results in a way that others can learn from, in a repository from which I will reap the eventual rewards of a solved problem, how can I begin to choose from among the 700 alternatives offered to me? How can I begin to focus on one, knowing at the outset that the effort may well be doomed at the outset and that, at the end of the process, I simply will not know which other combinations may have a greater chance of success. How can I even begin to figure out which combinations to use, when I have no sense of categories which to construct a partial ordering of the options?

Recommendations, anyone? eric

Sincerely,



Eric Armstrong
eric.armstrong@sun.com