Most widely used projections

[PeteD]

But if they show a scale bar applicable to the Equator at all latitudes, that, of course, is a mistake.

The length of the scale bar is adjusted correctly as you scroll up and down but can of course only be correct for one latitude, presumably the one halfway up the screen. This is fine at large scales but becomes more misleading the further you zoom out since the scale varies across the screen to a greater extent.

Perhaps the scale should disappear if you zoom out too far?There is after all a dedicated function for measuring distances.

[daan]

The length of the scale bar is adjusted correctly as you scroll up and down but can of course only be correct for one latitude, presumably tthe one halfway up the screen. This is fine at large scales but becomes more misleading the further you zoom out since the scale varies across the screen to a greater extent.

Perhaps the scale should disappear if you zoom out too far?There is after all a dedicated function for measuring distances.

I went though exactly this debate implementing a scale bar at my place of employment on Web Mercator maps. I argued that the scale bar should go away when you zoom out too far. The product manager decided that no, we can’t do that: it would confuse people. I said that they’re going to be confused about the scale if we show them a scale bar that’s grossly inaccurate over most of the map. My argument lost.

From which I learned, it’s better to confuse the human with misinformation than to confuse them with lack of information because at least they don’t know they’re confused with misinformation.

— daan

[quadibloc]

I went though exactly this debate implementing a scale bar at my place of employment on Web Mercator maps. I argued that the scale bar should go away when you zoom out too far. The product manager decided that no, we can’t do that: it would confuse people. I said that they’re going to be confused about the scale if we show them a scale bar that’s grossly inaccurate over most of the map. My argument lost.

From which I learned, it’s better to confuse the human with misinformation than to confuse them with lack of information because at least they don’t know they’re confused with misinformation.

I think I can be confident that this is sarcasm, and it's only "better" in the minds of managers with less integrity than persons such as yourself from a technical background.
However, I would tend to have some sympathy for the viewpoint that just making the scale disappear would make users of the product think it was broken. Mercator maps of the world often do come with scales that show the different scales at different latitudes in a two-dimensional form; a scale like that could pop in above a certain zoom out. Then the user would see that something is now different, preventing a simple scale from being useful, and so there would be no panic as a disappearing scale might lead to.

[daan]

I think I can be confident that this is sarcasm, and it's only "better" in the minds of managers with less integrity than persons such as yourself from a technical background.
However, I would tend to have some sympathy for the viewpoint that just making the scale disappear would make users of the product think it was broken.

You can tell I’m not much of a fan of “It’s okay to confuse people as long as they don’t know they’re confused” school of thought. To be fair, I did convince the product manager to let me put ~ in front of the scale value at medium and small scales and to put a tool-tip on the scale bar giving the ±accuracy for the scale bar. I think very few people would imagine there’s any tooltip to investigate, though, so I’m not convinced we did diligence.

The expanding scale bar that you note would not be permitted in this context because the screen real estate it would take up would compete with other things going on in the visualization that are considered higher value.

— daan

[Milo]

From which I learned, it’s better to confuse the human with misinformation than to confuse them with lack of information because at least they don’t know they’re confused with misinformation.

Ignorance is bliss.

Happy customers come back. Smart customers realize you're scamming them and leave. Given the choice, it's not hard to guess which one companies prefer.

[Milo]

...What if you change the scale bar's label from "Scale:" to "Scale at center:"? Only takes a few extra characters, not a lot of space, but it gets the point across.

[PeteD]

Like, temperature isn't a thing-per-area metric (in the sense that a 200-square-kilometer area isn't twice as hot as a 100-square-kilometer area, although you might argue it does have twice as much heat energy...), but it can still be useful to know whether it's better to think of Earth as mostly warm with some cold spots, or mostly cold with some warm spots.)

In the case of a temperature map, you might also want to know the rate of change of temperature with respect to latitude, which would be easier to visualize using a projection with equally spaced parallels than using an equal-area projection (except the sinusoidal, but then you might have other reasons for not using that). With a good compromise projection, areal distortion should still be low enough to tell that the Earth is mostly warm with some cold spots, although it could admittedly cause you to overestimate the size of those cold spots.

[PeteD]

Incidentally, you just proposed an application where the equirectangular projection would be favored Well, that or Eckert III/V, I guess.

Or any of:

• Wagner III, VI
• Kavrayskiy VII
• Francula I, II, VI, VII, X, XI
• Canters W06
• Winkel I, II
• Cabot
• Györffy A, B
• Apian II
• Ortelius oval
• loximuthal

In the case of a lenticular projection with equally spaced parallels along the central meridian, you'd still get the rate of change of temperature with respect to latitude along the central meridian, and as long as the graticule is shown, you could roughly visualize deviations from this rate of change at other longitudes. If that's acceptable, then you could also use any of:

• Wagner IX
• Francula III, IV, VIII, IX, XII, XIII
• Canters W09, W13, W14
• Gott-Wagner
• Tobias's autobiographical
• Winkel tripel
• one of the Winkel-inspired projections with rounded corners that I presented in the "Experimental projections" thread in April
• Aïtoff

I've deliberately omitted the azimuthal equal-area, American polyconic and rectangular polyconic projections because the parallels curve far too strongly beyond the inner hemisphere. Have I missed any others?

[quadibloc]

...What if you change the scale bar's label from "Scale:" to "Scale at center:"? Only takes a few extra characters, not a lot of space, but it gets the point across.

If you've zoomed out so far that, say, 10 degrees South Latitude is visible in the map, which is on the Mercator projection, but the center point of the area you're looking at is at, say, 50 degrees North Latitude, then that wouldn't quite be accurate. The Equator would not be anywhere near the center of the map.

[PeteD]

But the length of the scale bar is adjusted to show the scale halfway up the screen, not at the equator, so "scale at centre" would be accurate.