Last modified: |
11/1/04; 5:43:00 PM
|
Feeds: |
LIVE webcam Cumbres & Toltec rail yard in Chama, New Mexico.

Current BlogRoll.




[Macro error: The server, api.google.com, returned a SOAP-ENV:Server fault: Exception from service object: Invalid authorization key:]

|
|
Saturday, October 9, 2004
Microbatteries Built on a Bed of Nails [ Slashdot:]
< 12:09:38 PM
>
Broadband over Power Line communications appears to interfere with "RFID" tags used to identify inventory and other items.
Next week, on October 14th, the FCC will also announce its Report & Order regarding
power line communication systems. Recently, the National
Telecommunications and Information Adminstration (NTIA) has issued
statements now generally supportive of BPL, even though the NTIA itself
documented radio interference problems in the past.
In spite of its positive spin on the interference, the NTIA made a filing with the FCC on September 13, 2004,
(PDF document) in which it requests a complete prohibition on BPL on
frequencies allocated for aeronautical radio in the 1.7 to 30 Mhz
band and radio navigation in the 74.8 to 75.2 Mhz band (see page 16).
The NTIA found that when many BPL access devices operate simultaneously
(as would be seen from an overhead aircraft), significant communication
degradation occurs.
The NTIA also requests that BPL be prohibited within 1 km of all
coastal (e.g. ship and aircraft) communications stations (page 18).
NTIA requests that BPL be prohibited within 80 km of radio astronomy facilities.
BPL systems within 4 km of FCC field offices and other "protected"
locations must be coordinated between the power company and the
organization, and should coordinate BPL within 80 km of certain radar
receiver systems.
This document represents a rather stunning confirmation of the
interference problems associated with BPL. Yet the FCC is hell bent on
approving BPL on the basis that all interference must either be
tolerated or can be magically "coordinated".
Meanwhile, yet another utility has backed away from BPL.
Also next week, my own local electric utility has scheduled a meeting
with the Amateur Radio community to discuss BPL options in the Spokane,
WA/Couer D'Alene, Idaho markets. They say they wish to meet with
Amateur Radio operators before hand and work to resolve issues before
they begin a BPL test system in this area. The meeting will be held Oct
14th at 7pm in the Avista headquarters building at Mission and Upriver,
in Spokane, WA. (Noting that the date is the same as the FCC's R&O
date, one must wonder if the utility has been tipped off as to the
content of the R&O?) I hope to attend, as representative of the Red
Cross.
We use both amateur and non-amateur communications, including systems
on frequencies that would likely not be protected from RF interference
generated by BPL systems. [Edward Mitchell: Common Sense Technology]
< 12:08:52 PM
>
Associated Press caught making
up the news, again. As I noted last week, many election polls indicate
a close election, so close that the only valid statistical conclusion
is that neither candidate can be judged ahead of the other. Early
this week, MSNBC got it right and called the race a "statisical dead
heat". But other reporters are, unfortunatley, either somewhat
dimwitted on elementary statistics, or they are outright making up the
news stories.
The last couple of weeks, Newsweek boasted of a "tightening" race, when
in fact, the confidence interval of the survey was so wide they could
not conclude anything. Now, in AP's latest poll,
Kerry is proclaimed as having a "slight lead". In fact, the data does
not support that conclusion either - AP made up the news report
to say that.
Once again, here is a simple explanation of a poll or survey. A sample
of potential voters is surveyed to learn their voting preference. (Only
7% of those contacted even agree to participate in a poll, I've read. I
was actually called a week ago and declined to participate.)
This produces an estimate of the population mean. For example, 50% for
one candidate and 48% for the other. This however, is only a guess of
how the entire population will make their selection.
When we expand the sample to an estimate for the entire population
of all voters, we introduce a margin for error. That is, because we
have only sampled a small number (relative to the population), we might
have sampled a biased group and our survey results could be wrong.
From statistical theory, a confidence interval can be calculated giving
us an idea of a likely error range for the estimate we make for the
entire population, based on a small sample.
For example, we may have calculated a standard error of + or - 4% in our survey result. What that means is that - if we had measured the entire population instead of a small sample,
the genuine population result will fall within the standard error or
confidence interval or, for this example, 46% to 54% and 44% to 52%.
Since the actual population result could be anywhere within these ranges, we have no way of knowing who is ahead.
Every time a "journalist" reports yet another survey result with
overlapping confidence intervals and tells us the race is tightening,
they are publicly proclaiming that they have made up the news.
This is very unfortunate and leads to genuine mistrust of news
reporters and editors.
I won't write much about this during this election cycle - except to
point further examples. Its just so frustrating to see such shodd work
being passed off in news and headlines, daily. [Edward Mitchell: Common Sense Technology]
< 9:05:22 AM
>
Wow! CNN writes "an initial poll indicates a statistical tie in Friday night's face-off". Read it. CNN gets it exactly right and explains why. Finally! [Edward Mitchell: Common Sense Technology]
< 9:04:00 AM
>
|
|
|