On Sunday, April 4, 2004, at 09:53 pm, Alejandro Dubrovsky wrote:
100%: Original Star Wars rereleased in theatres
<
This supposedly wasn't a wrong prediction. It got judged at a 100 (it
paid one cent per million dollars in tickets in the US (up to 100 off
course))
OK. I see how to read the judgments now. That will help a lot in
interpreting these!
But I am still confused how the original Star Wars was judged 100% to
have been re-released *in theaters*. It says
"TimesIf the Star Wars
movie is not re-released to theatres by 1/1/99, the claim pays $0."
Was there a theater re-release of the original Star Wars that I
missed? I thought they just released a DVD.
100%: MiniDisks will outsell CDs
<
Check the graph, some weird anomaly where it jumped from 2% to about
100 and back instantly. some kind of interface screwup most likely.
If the data graphs are not trustworthy, it throws my entire attempt at
historical analysis out the window!
100%: Terrorists have nuclear weapons
<
It was at 100% for less than a day at the start. Have a look at the
graph.
I know. This doesn't invalidate my claim. "Predictions" that seem
100% obvious at first can quickly cool off and be thought of as silly
later. This prediction still peaked over 60% several times. It
clearly was a common belief, and many people still expect terrorists
to get nuclear weapons any time now.
100%: Hillary Clinton runs for president
<
Correct (don't read the title, read the claim). The actual claim is
that she would run for some position int the 2000 election, which did
occur, and it got judged at 100.
Ouch. So the title are sometimes inaccurate, too? It is really hard
to use this "data" with all these inaccuracies that we have to ignore.
I admit that I quickly scanned the titles, dates and graphs. I didn't
realize that any of them could be wrong or superseded by other data
elsewhere in the claim text.
100%: Non-Intel PCs dominate market
<
Another one where it looks like an interface screwup. Look at the
graph. It jumps from about 35 to 100 and back instantly (twice
though).
How can people quote this stuff as authoritative for their claims, but
then ignore it for opposing claims? If the graphs aren't reliable,
why trust any of them?
90%: Colin Powell will run for President
<
Real top value about 80%, but yes, wrong prediction.
90%: Unix irrelevent
<
So is this one
Even with your other points about anomalies or inaccurate titles,
there are still numerous claims like these that were widely supported
and just outright wrong. My point that these predictions don't prove
anything still stands. Widespread market consensus is not good
evidence that the prediction must be right.
85%: Mac clone sales outnumber Mac sales
<
Only right at the start, quickly drops
Again, this supports my point. Predictions that are widely believed
to be "obvious" at the time quickly disappear after a little analysis.
(Unless your point is that these initial peaks are anomalies as well.)
80%: Netscape wins browser war
<
Read the claim. It got judged at 54, so 80% at top is not bad.
You can go through the rest at your own time. Main point: don't read
the title, read the details, and look at the graph
Yes, I admit that I did skim these very quickly and expected the
title, date and graph to be accurate all the time. It now appears to
me that they are full of errors or anomalies, and only the text and
final judgment can be trusted(?). Even so, you refuted less than half
of my examples. There are still many obviously wrong ones. All I am
saying is that these predictions are often wrong.
--
Harvey Newstrom, CISSP, CISA, CISM, IAM, IBMCP, GSEC
Certified IS Security Pro, Certified IS Auditor, Certified InfoSec
Manager,
NSA Certified Assessor, IBM Certified Consultant, SANS Certified GIAC
< <