BIGEarl said:
The basic problem that I see is inconsistent data. Accuracy is not the issue.
Sure, consistency across sources would be nice. But bad data is still bad data even if you get it from several sources. (And actually, if it is one original source and passed to you by several messengers, it is still one source and any confidence that you get from the multiple consistent messengers is false.)
And I would argue that accuracy is also important. If one trail is listed as being 20% longer than a realistic value and another is listed as being 30% shorter it would lead to unrealistic expectations of the difficulty of hiking one trail based upon the difficulty of hiking another. (Actually this is an argument for consistent error factors, not accuracy. I'd add that, IMO, the easiest way to achieve consistent error factors is accuracy.)
Look at the definition of book [hiking] time. It is well defined and easy to understand. Most of us don't hike at that speed, but it is a reasonable definition and many of us know what fudge factor to apply to estimate our own hiking times. (Ie they are consistent. There is no general ground truth here so there is no precise notion of accuracy. But in a general sense, the times are in the ballpark for most hikers.)
An analogous convention for measuring hiking distance would also be appropriate. The old method of using a surveyor's wheel of specified diameter is a practical method that is presumably consistent and reasonably close to a hiker's perceived distance (ie accurate). It has the problem that someone has to walk the trail dragging the apparatus and it is only practical to publish distances between points-of-interest (POI, eg trailheads, junctions, and summits).
The White Mountain Guide is offered in two versions, a printed book and an on-line service. Since both products carry the same name, are intended to serve essentially the same purpose, and are provided by the same organization it seems reasonable to expect the information provided on any given trail segment to agree from one source to the other. This is not currently the case.
The online user guide does inform the user that there are differences--the book distances are measured by a surveyor's wheel and the online distances are estimated by an algorithm. There is a similar statement regarding the differences in the total climb for a route.
Comments in the user documentation seem to indicate the online information is believed to be more accurate than the printed information. This isn’t my opinion but the opinion of the organization promoting the products. They are evidently discrediting one product in favor of the other. That’s an interesting position. Are they selling books, subscriptions to online services, or attempting to stimulate an internal war? Perhaps there’s some entertainment on the way.
I saw nothing in the online docs available to the general public that indicated that one was believed to be more accurate than the other--just that they were different. Could you point me to such a statement?
Is this the opinion of a cartographer, someone involved in producing or promoting the online product, or just some random individual who was tasked to write the docs and/or advertisements. Did they bother to ground-truth the numbers generated by the online product? Unless the online numbers were ground-truthed by cartographers and people who understand how to measure distances on digital routes on digital maps, I'd be inclined to assign a low confidence to such an opinion.
Here is a link to an article giving an analysis of the errors in distance computation in NG TOPO!:
http://tchester.org/sgm/analysis/trails/chantry/topo.html. FWIW, NG TOPO! Northeast has often been observed to generally give shorter trail distances than the WMG book.
FWIW, I can think of distance measuring algorithms for routes on maps whose errors depend upon the general heading of a route and the method of inputting the route.
Pragmatically, it is easier and cheaper to provide an algorithm to measure distance on digital maps than to maintain a table of distances. (An algorithm can also give a distance from anywhere to anywhere by any route which is impractical to do by a table.) But there are a zillion* pitfalls in applying such algorithms to real routes on real maps. While the general user may not be aware of the details, he should be assured that there has been appropriate professional oversight and validation. Or at least something describing the errors and their sizes.
* Well, maybe only a half a zillion. But a lot, anyway...
Granted, the book is as good as it has ever been. Unfortunately, it can be improved but there seems to be no plan to make the needed changes.
As noted above, it isn't clear to me that the online distances are superior to the book distances. The book distances are measured by a method whose errors are fairly easy to understand (and perhaps quantify), not necessarily the same for the online version.
Doug
Makes ya wonder doesn't it.
