Home | Contents | Search Iago Conspirators Timeline Junot
From: Jasper
Date: 4/3/02
Time: 1:45:58 PM
Rovaan,
You can't begin to imagine how stunned I was to see the featured article on Wagner's site called "Heidstra's Timeline" (I thought I made that up) and some of the methodology he used, which was eerily similar to mine.
He does, however, suffer from an analytical weakness that gets young engineers in big trouble when they try to build an accurate model from incomplete data. They assign precise values to speculative variables then proceed to add the variables and tally the results. It looks impressive but it doesn't work.
The fatal flaw in the "young engineer approach" is that tiny errors in the data and the operations are cumulative. Given enough operations the errors invariably overtake the correct information and the solution becomes about as useful as taking a big guess in the first place.
The problem is analogous to measuring a long table one inch at a time by first taking your best guess at what an inch is then using each newly measure inch for the next measurement until you're finished measuring the table. Even if you take a known portion of an inch and guess at the rest, each tiny error builds on another until the end result is useless.
To solve a problem like this, you have to eliminate as many variables as possible, reduce the ones you have to use the simplest terms and apply them with the fewest number of steps. When you start seeing fractions in the early stages of the process you know you're in trouble.
The key to modeling the Heidstra timeline is to find the start, the finish, the constants, the deviations from the constants and apply as many objective, independent checks as necessary to see how well they match up. Wagner did a pretty good job with the checks but Shively is still a handicap.
Average speed, precise distances per se, and variations in the terrain are irrelevant. What's important about the distance (whatever it is) is that it can be proportioned in manageable increments that average out the variables in whole numbers within each increment. Working from any known point in time and space forward or backward to any unknown point within the model will give you an overall margin of error that cannot exceed a fraction of the nearest known point.
The photos and drawing of the Heidstra rectangle are far superior to anything I've ever seen and anything that I've been able to come up with using the information at my disposal. They would be extremely useful in this discussion. I urge everyone to check them out. I can see that Wagner's site offers many things that would be of value here (whether we agree with his reasoning in all cases or not) and I will add his sight to my links right below the Walraven transcripts the first chance I get.
Thank your Rovaan for another magnificent contribution. --Jasper