[RRE]Wish List

From: Phil Agre (pagreat_private)
Date: Sat Nov 10 2001 - 09:43:06 PST

  • Next message: Phil Agre: "[RRE]Summary: Carlyle Group and Bush Administration"

    I received many responses to my "Wish List" article of October 20th.
    This message excerpts some of the responses that might be useful to
    others.  I have heavily edited them, both for length and to suppress
    people's identities.
    
    I have refrained from responding, with one exception: the person whose
    comment on wish list item (10), "a book that explains the adult world
    to children", persuaded me that my analysis was obnoxiously screwed
    up.  My intended meaning was this: scaffolding by parents and teachers
    is already one of the major ways that children learn, and we should
    invent artifacts that amplify the effect, with the goal of giving kids
    a clearer participatory understanding of the adult world.  The point,
    again, is not that parents aren't raising their kids the right way,
    but that with new cultural artifacts we can do better.  The analytical
    error in my piece was one that I often warn against here: I presented
    the new technology as supplying something that is missing in the
    world, when in reality it would be amplifying something that already
    happens.
    
    Finally, a general comment that applies to several of the ten items.
    I should have made clearer that my "wish list" looks forward to the
    expected state of technology ten years from now, not what we can do
    today.
    
    
    (1) Personal Access Guide.
    
    Yvan Leclerc of SRI has invented the basis for your Personal
    Access Guide.  It's called GeoWeb and you can read more about it
    at http://www.geoweb.net/ and http://www.geoweb.org/.  GeoWeb uses
    a global meta-directory to call up online information relevant to
    any point on the globe.  It provides filters to produce just the
    information that you're looking for (assuming it exists on the Web).
    Unfortunately, the .geo TLD was not approved by ICANN.
    
    
    Sheffield Green Map
    http://www.shu.ac.uk/schools/cs/greenmap/
    
    
    (2) Real-time water quality indications at beaches (and automatic
        sensing of other invisible things).
    
    Southern California Beach Report Card: 2001 Summer In Review
    http://www.healthebay.org/brc/2001summer/default.asp
    
    Studies Find Many More Sources of Coastal Pollution
    http://www.latimes.com/news/printedition/california/la-000084899oct25.story
    
    
    For five years (1996-2001), EPA program funded a program called EMPACT
    (Environmental Monitoring for Public Access and Community Tracking).
    Here's the website: http://www.epa.gov/empact/.  They are no longer
    giving funds to try new methods (thank you Mr. Bush) but they are
    promoting the spread of the technology.
    
    The program is supposed to promote both technology for real-time
    information and community organizing to interpret that information.
    The program has wrestled with some degree of success in finding ways
    to resolve the problem of different expectations for information.
    
    The program in Boston made a difference here b/c it allowed organized
    local folks, the Charles River Watershed Association (www.crwa.org),
    to pay for the installation of a system to monitor the river in real
    time.
    
    Real-time information about bacteria counts is currently not feasible
    because the best method to determine the amount of bacteria in
    water is to culture a sample of water in a lab.  This takes 24 hours,
    trained staff and controlled conditions.  The City of Somerville,
    MA; the Mystic River Watershed Association and Tufts University
    have received one of the very last grants from EPA EMPACT to test
    a model that would use a statistical analysis of several parameters
    to estimate the bacteria counts real-time.  It will be in operation
    in 2002.
    
    
    You asked for a "thermometer" that tells what is happening in the
    electromagnetic spectrum in a given place.  Here's what NTIA uses
    for that purpose:
    
    http://www.its.bldrdoc.gov/home/programs/rsms/  
    
    There are "suitcase" versions but they're limited in the frequency
    range and level of emission that they measure.
    
    
    (3) Self-diagnosing public address equipment (and Internet gear).
    
    Color coding is a huge step forward as far as computer cables go,
    and it's becoming standard.  It would be a huge help and a good start
    if people would actually READ the instructions that come with their
    computer.
    
    Also, there is free network monitoring software that's available for
    both linux and windows that monitors the 'services' on each server.
    I'm most familiar with the liinux software, which is called NetSaint.
    It monitors ever single server we have in the building... if one goes
    down, someone is paged within five minutes.  It also monitors each
    service, such as the Apache (website) server on our intranet machine,
    and lets someone know if that particular piece of software crashes.
    
    As for audio equipment... 8 times out of 10, a break in A/V equipment
    was not preventable ahead of time (A fuse blew, a tube blew, a board
    or chip wiggled out of place, something overheated...) ... 1 time out
    of 10 it was caused by a presenter or rock star who thought "something
    didn't look right" and "fixed it"...  And the other time out of 10,
    it's shoddy testing ahead of time.
    
    
    With regard to the suggestion that Internet gear should diagnose
    itself, there are two separate problems here: diagnosing the network,
    and diagnosing the last-mile connection to the end users.
    
    All ISPs I've dealt with have extensive infrastructure for proactive
    monitoring and fault detection in their networks.  They're often
    the first to know about things like gas-line explosions and train
    derailments because the network monitoring stations detect the cable
    cuts that are associalted.
    
    At the large ISP where I worked, I would often inform my customers
    that they had a network problem by calling them to tell them it was
    fixed.  Our monitoring software quickly and automatically informed
    the operations staff about broken links, excessive bandwidth usage,
    unreachable hosts, and so on.  I'm sure it has since been augmented
    to include various security issues (e.g., denial-of-service attacks).
    
    Consider, on the other hand, an ISP with only 200,000 customers.  Even
    if cable modems are only $100 each (a ridicuously low estimate) that's
    $20,000,000 in installed hardware. Add $10 for self-diagnostics (again
    ridiculously low); add $2,000,000 dollars in capital investment.
    
    Now you start monitoring individual customer links.  How do you tell
    the difference between:
    
      - modem powered off (the most common state for a cable modem)
      - computer powered off or unplugged (the second most common state)
      - one of various cables unplugged
      - network link problems (the one you're interested in?)
    
    in only $10 worth of hardware?
    
    Even with the expensive gear, I tell that a customer was down, but I
    could not tell *why* without feedback from the remote (impossible when
    the link to the remote is down :).  I'd have to pick up the phone and
    talk to someone.
    
    Self-diagnosis is expensive; we've already spent the last 20 years
    proving that people are willing to sacrifice quality for a cheaper
    product, *especially* in an environment where that product is obsolete
    long before it breaks.  And I'm not sure it would actually improve the
    experience.
    
    
    (4) "Whatever happened to ...?"
    
    I work for a consulting firm that makes forecasts within a broad
    category of technology.  Sometimes we are covering highly speculative,
    "emerging" markets that, although not strictly in the "New Economy,"
    often have that flavor.  Proponents of these technologies wax
    enthusiastic about their impending hockey-stick growth curves, which
    always start their vertical ascent about 4-6 quarters in the future.
    
    My company takes an unusual attitude toward market forecasting: we
    try to get the right number.  While I share your frustration with
    companies like Whatever-You-Wanna-Hear Consulting Group (one of our
    main competitors), I am also well aware that the situation in the
    forecasting world is complex.
    
    The suggestion, "Why not go back and look at the forecast numbers,
    compare them to reality, and hash out the differences?" is in some
    sense what we try to do in our research reports, although mostly we
    are simply explaining why last quarter's/year's numbers in our own
    reports changed a bit due to new market conditions.  We also do it
    internally as a measure of our company's accuracy.
    
    The reasons it is not done in a broad sense are multiple and perhaps
    overlapping.  Let me just toss out some reasons:
    
    1. It is not especially desired.  Seriously.  Businesspeople don't
    care much about the past, which is over and done with -- the money and
    time already spent.  We have thought about publishing a retrospective
    that shows our long history of decent prediction, but in reality, few
    clients would read it.  All eyes are on the future.
    
    2. The purpose of market forecasts is not only to get the numbers
    right.  (We stick to that adamantly, but we are realistic enough to
    know that market forecasts serve other purposes too).  For instance,
    they are used to garner investment dollars.  Young companies writing
    business plans need big numbers for the VCs.  VCs are rich but not
    necessarily smart, and may simply believe the entrepreneurs.
    
    3. Also, there is a feedback loop that operates through the media,
    consulting firms, and their forecasts.  Firms that have big numbers
    get quoted more often, which means their name gets known, so they
    get more calls from journalists asking for numbers.  This tends to
    propagate the larger numbers through the public awareness.  (We have
    noticed that many firms buy research reports both from us and our
    competitors.  The inflated numbers show up in press releases, while
    ours are used for internal planning).
    
    4. Remember also that the "market size" is a fairly slippery concept
    anyway.  Yes, there really is some exact number of widgets bought and
    sold each quarter for a certain average price, but this number is in
    fact unknowable.  Why?  Because the only way to find it out is to have
    an independent researcher, like someone from my company, call around
    and ask the players how many they made or sold.  And you know what?
    They lie.  Companies know that these data will lead to pie charts of
    market share, so they inflate how many they sold.  (Although there
    are also circumstances where they downplay how many they sold to
    make markets look smaller, such as when their competitors are ramping
    up in that area).  Our job is to sort through the baloney and glean
    some sort of best estimate.  Also, we cannot sample the whole market,
    so we have to guess after polling some fraction of the total players.
    
    5. Finally -- now really entering the realm of the non-scientific
    -- market forecasting contains some of the same elements that drive
    the stock market.  Markets are partly created by people's perception
    of them.  I give this notion far less credence than most people
    in my field, but I grudgingly give it a little.  Big forecasts can
    lead to more sales and vice versa simply through emotional reaction.
    This will *never* create a long-term trend, but it can have a small
    effect in a given quarter.  (This is related to number 2, also, to
    the degree that the presence of VC funding in a technology area feeds
    the perception that it is a "hot" area.  This then attracts talented
    workers looking for career boosts, etc.)
    
    Given all of this, it is still understandable to want follow-up
    checks.  I don't think Whatever-You-Wanna-Hear Consulting Group
    should get away with playing the media and basically buying business
    by currying favor.  But the idealistic concept of using this data to
    look for patterns in our prediction methodology makes little sense in
    light of the reality of the business.  Few forecasting companies had
    accuracy in mind in the first place, plus there are the confounding
    variables of prevarication in reporting sales and emotional effects
    in buying patterns.  Furthermore, I predict that the market for such
    information is relatively small :-).
    
    
    (5) Automatic evaluation of ISPs' spam-enforcement efficiency.
    
    http://www.spamcop.net/
    
    makes reporting spam easier.
    
    And here are their stats pages:
    
    http://spamcop.net/spamstats.shtml
    
    
    (6) Consumer Reports for design.
    
    Adbusters has run some campaigns aimed at reinventing design in a more
    responsible fashion:
    
    http://www.adbusters.org/campaigns/first/
    http://www.adbusters.org/campaigns/first/tour/2.html
    http://www.adbusters.org/magazine/36/toxic/
    http://www.adbusters.org/magazine/36/next/
    
    Most of these pages have more of the article on the following pages.
    I still think the ideas behind the First Things First manifesto are
    good.
    
    
    (7) Daily reports on party lines.
    
    This is perhaps what National Journal Hotline is...
    http://nationaljournal.com/pubs/hotline/
    
    
    (8) Online music sites that tell you how long each recording is.
    
    http://www.cddb.org/ is indexed by track/album length; that's how the
    automatic lookup algorithm works.
    
    
    (9) Shopping basket price comparison for online commerce sites.
    
    I haven't used it for some time, so my memory may exaggerate its
    ability, but I think that ABEBooks has/had an engine that would
    compare prices for bundles of books.  I'm guessing that their system
    relied on used book sellers sending lists of books and prices to them
    for combination into a big data base.  It's not really what you asked
    for, but perhaps a step for some types of products
    
    
    Something close to what you want used to exist.  In 1998 and 1999,
    I used a website (it no longer exists -- see below) that would let
    you create a shopping cart full of stuff you wanted (limited to books,
    music cds, and videos, but at the time IIRC Amazon didn't sell much
    more than that) and then would query a bunch of major online stores to
    give you the best price, shipping included (you gave it your zip code
    so it could figure that out).  In later versions, it would even figure
    out if partial orders from multiple stores would be cheaper than
    buying everything from one.  Then, if you were happy with a particular
    offering, it had a link to the site(s) so you could directly place
    your order.  Alas, it's interface was somewhat clunky requiring
    multiple steps to get something into your "shopping cart", but the
    information it gave was too useful for me to let that get in the way.
    
    Unfortunately, it got bought and morphed into several other web sites,
    and eventually turned into something much less useful before vanishing
    completely as far as I can tell.  (I haven't mentioned the URL because
    it now links to a nasty domain name squatter whose site will attack
    your browser.)
    
    Here is a FAQ on the service dating from late 1998:
    
    http://web.archive.org/web/19981207051937/www.acses.com/faq.htm
    
    and also this help tip on pricing multiple books at once:
    
    http://web.archive.org/web/19981207005921/www.acses.com/help.htm#Multiple
    
    
    (10) A book that explains the adult world to children.
    
    I don't know about the children you spend time with, but let me say
    that as a parent, I spend a large amount of time doing the activities
    you mention and more.  And I think this is true for most parents I
    know.  In the past few months my son, who just turned two last week,
    started to demand handling the money in purchasing transactions, and
    both my wife and myself let him, and guide him through it.  And often
    in coming to a place, I spend time explaining to my daughter why we're
    there, what I'm trying to do, and how I go about doing it.  These are
    just a few of the examples of what my wife and I do to bring up our
    children into adult world.  And most parents I know do the same.  And
    I'm not even mentioning all of the role playing we do around these
    activities.
    
    end
    



    This archive was generated by hypermail 2b30 : Sat Nov 10 2001 - 10:45:36 PST