- Support #58
closedrevise the bezier curve for the ERA paper
0%
Description
While working on the text update, I notice that the curve in version 0.6.* of the paper suggests that ER Index of 10 on the logarithmic scale corresponds by general design to the Worst Scenario, but that's not the design intent. There is not limit of log value 10. I attach an illustration with a curve that hits the same x and y samplings, but that runs much higher than the log value 10. This is an important fix to the illustration for the reader to obtain a correct understanding of the design.
Files
Updated by Joseph Potvin over 2 years ago
As I work on the text, my brain also returns to the preference to have the diagram EITHER:
(a) in a vertical configuration, as in the rough sample attached.
OR
(b) if in the horizontal, then with the Bezier portion flipped the other way so that a long tail on the "worst" side is easy to extend
Either of the attached makes more sense to me when talking about the mechanism rationale.
I reckon the NEW horizontal works best because the Bezier now shows "progress" going up and to the right, which is conventional in charts.
Updated by Joseph Potvin over 2 years ago
- File Adjusment-to-ER-Index-chart_horizontalPNG.png Adjusment-to-ER-Index-chart_horizontalPNG.png added
Addition the horizontal variant. But this does not make sense because the x-axis for the curve would have to be backward oriented (the larger numbers further left).
So I conclude that the vertical one is optimal.
Updated by Joseph Potvin over 2 years ago
In discussion yesterday with Jacob, he raise the legitimate issue that if the Bezier spline is (in effect) asymptotic to the "Worse Case Scenario", then some natural or human-caused event which pushed close to that worst case would, in effect, crash the system.
In my reply yesterday I forgot to mention how this can be handled in a working ERA system. I did say correctly that the curve is designed to make approaching the Worst Case untenable. I should ALSO have said that in the even of a disaster (Fukushima tidal wave; Mount St Helens blast; or human-caused catastrophes such as the dike maintenance failures in New Orleans and Abbotsford), in the ERA system, the stakeholders can reset the base year. The precedent for base year resets is in the CPI (consumer price index) realm -- certainly this is occasionally abused. So a governance mechanism over the methodology is required. But generally speaking this is how the asymptotic curve is prevented from crashing the system.
Updated by Calvin Hutcheon almost 2 years ago
Is this still an issue that needs to be resolved?