An efficiency frontier of mean and standard deviation could be plotted where each point is a place name (or desc or address or whoever). The x coordinate of the point representing a place id in the database is the distance between its coordinates and the coordinate-equivalent of visitor inputs. The
y coordinates of points in efficiency plots represent the vector standard deviation of the point as represented in the database. For example, if the information only contains city and state, the y coordinate would be the rotational moment of inertia of a cutout of that city from a paper map, and the x coordinate would be the distance from the coordinates represented by the centroyd of the city map cutout and the coordinates representing visitor inputs. The most efficient hull would be the southwest hull, not the northwest hull as is considered efficient in financial portfolios. We're talking about geography, not economics. The resultant markup that is the function of user inputs would logically include a table whose left column is an ordinal number (1, 2, 3, 4...) representing efficiency hulls progressing from southwest to northeast. The rightside column would be the number of database items along each hull. For forms enabled visitors there would be a check box for whole enchilada (might as well free information iff that's what it wants) with top 10 or so by distance or some other criterion being the default cgi output.

Please explain more, or give examples. What sort of data are you planning to represent? Is data with a high mean and low standard deviation as good as data with a low mean and a high standard deviation? I see this might be the case with distances, where nearer is better, but does it apply to other data sets? And for what purpose would you use this? (I wasn't certain if the x-axis is always meant to represent distance, or if that is just the example given.)

As I understand, the central principle of a GIS is that every data value has a 2-D or 3-D location attached to it, allowing data from different maps, at different scales or with different projections, over different areas, to be integrated and compared. Clearly being able to locate data values that are geographically near to a given point is important. I believe this idea is aiming to relate the location of data measurements to the accuracy of the data, I just can't see exactly how it would work (which isn't to say that it wouldn't work).

Geometric note for other readers: the centroid of a 2-D figure (such as a state outline) is the centre of gravity, and can be taken as the centre of that state. The rotational moment of inertia about some axis is the sum of (mass times radius squared) for each point: I don't known why LoriZ doesn't just use the area (is she mapping mass distribution?)

So, a travel location that tops the 'well kept secret' list would have a relatively low standard deviation like the low, more stable ß of financial tables. The location would also register a broad array of user-derived efficiencies due to the characteristic of offering a little something for everyone. There is a case for using the median measure for ranking instead of the mean, just to leave the locations in perspective.

There used to be a web site that did this. Can't recall the name (thought it was SETI@Home, but I can't find anything there now) but I'm pretty sure it was a science-oriented site. Perhaps something detailing Internet usage. If I find it I'll post a link.

Near as I can tell, this is a service for identifying geographical locations which match a user's criteria (distance from a certain point, rotational inertia?).

I'm not clear why anyone would care about the rotational inertia of a city, or the purpose for which this is intended (finding a place to live? trying to locate a place you know about but can't recall the name of?). Nevertheless, I don't think it's Vernon; it's unclear but it's not nearly as much nonsense as it seems.

"the centroid of a 2-D figure (such as a state outline) is the centre of gravity, and can be taken as the centre of that state".... reminds me of the description of the 19th century mapping of Minnesota, and the reason that Lake Wobegon was missed out. Four surveyors started at the corners of the state , mapping around them as they went towards the middle, where they joined up their maps (they had all missed that little town where all the men are strong, all the women are good-looking, and all the children are above average.)

I did not post this Idea, and I'm not certain as to its purpose. A few things about it, however, lead me to speculate....

LoriZ appears to be describing a graph that will contain multitudes of data points, each representative of a certain degree of accuracy-of-information (fuzziness). Next, just as in a simple graph we connect related data points with a single (often jagged) line, in a complicated graph like this we could reasonably desire to connect only data points of equivalent fuzziness. The result would be a multi-line graph, and possibly the lines would have similar contours. ("Contour" may be a synonym of "hull", but I'm not betting on the accuracy of that interpretation.)

Re. "as good as"...in EFA as I (mis)understand it, no assumption is made as to which "compromise" is better. Re. Iff...LoriZ doesn't use if and iff interchangeably. Re. moments...having thought about it, area makes much more practical and intuitive sense than moments. Except maybe for "gerrymander analysis." Reensure...Re. medians...PLEASE tell me more! Efficiency hull, try replacing "naval architecture" with "optimi(s/z)ation", "convex set", "nondominated", "Pareto" etc. LoriZ is not Vernon. LoriZ is not LoriZ, either. LoriZ is a "handle". LoriZ is Lorraine Lee from Detroit. LoriZ was the original N8CHZ, but that callsign is retired (i.e. expired). Last I checked it was not in use. You (anyone) could be the next N8CHZ if you live in America and so desire. Thank you (pl.) for your support! LoriZ contemplates the question of what would be the rotational moment of inertia of the states of Michigan, Ohio and West Virginia, combined. The general idea is how to deal with fragmentary geographic information, which takes many forms, not only of information, but of incompleteness. There's also of course the usual LoriZ WIBNI of somehow separating "informationally sophisticated" from "well kept secret."

In mapping software, the "points of interest" database could use this. For instance, we know the location of the McDonald's off the next exit pretty exactly. But we don't know the specific boundaries of WPHS's radio signal. Maybe it overlaps with WEMU, and if you're on the boundary line, the software should show you both when you ask what's on 89.1FM.

After marking waypoints with my GPS, maybe I want to measure the distance bettween them. But the receiver isn't perfectly accurate. So each point should also store a confidence figure or an error estimate, then any calculations done on them could include the same data. "The area described by these 3 points on a triangle is probably about 50 square meters, but it could be anywhere from 27 to 80, based on the error of the measurements."

My master's thesis touched on some of what I think this is about. There is no generally-agreed definition of a wetland, except as an area with a fuzzy boundary that changes with time, but computer maps give the appearance of accuracy out to a dozen decimal places. So I recommended old-fashioned methods of estimating wetland area, with emphasis on estimate.

But I don't understand a damn thing about this idea, except that "centroid" is spelled wrong.

[Myself], GPS points are very rarely given any estimation of error, but they should be. When 'tis done, it's with a CEP diameter, which is worse than it sounds. Circular Error Probability estimates that the given point has a fifty-percent chance of being within that size of a circle around the actual location. Which leaves out vertical error, and does not address the level of confidence in the CEP estimate.

If that doesn't make sense, don't feel lonely. I sat through a two-hour high-level presentation about GPS surveying, at which there was no mention of the possibility of error. The guy beside me was very upset by that lack, but couldn't get anyone to take him seriously.

[baconbrain], two things. First, regarding wetlands. I've done a fair amount of consulting involving wetlands and what most people miss out on is that a "wetland" absolutely does have a hard boundary.

I say that because a wetland, in any sense where money is going to change hands is purely a legal concept and a line gets drawn on a map. There it is... it gets signed off by the Corps (or whoever has jurisdiction) and there it lies... for that day -- for that transaction.

A lot of resources, natural and otherwise, could be saved if the various agencies were better able to quickly agree on exactly what does constitute a wetland. Remember, even though the existence of a "wetland" is a natural phenomenon that is characterized by various wetland vegetation and other things, it is primarily a legal concept.

Second, surveying GPS is done with differential correction and the actual accuracy is usually less than a centimeter. That's probably why the discussion of possible error was eliminated from the 2 hour presentation you saw... it's because no one really cares anymore.

//a wetland, in any sense where money is going to change hands is purely a legal concept //

That's a *very* good way to look at it. Thanks. But don't tell my thesis advisors or they make me go back to school.

My initial approach to wetlands was within the National Park Service, where wetlands were not going to be sold. And I leveraged that experience into my thesis.

Regarding GPS surveying: You are right there, too. I helped set up a differential base station while in college, but nobody uses it anymore, so we shut it down. The lecture that I referred to took place before the accuracy increase, though.