html xmlns="" xml:lang="en" lang="en"> From the archives: I've been thinking about this slide for days now.

Monday, September 17, 2007

I've been thinking about this slide for days now.

I do not like this model display one bit. Oh no I do not. It makes me ANGRY, because it talks a good game and promises a whole lot, but then it plays on our subconscious assumptions to LIE TO US. It is a big fat lying triangle, and I do not appreciate being LIED TO by geometry. Further, it is actually a display that could have a lot of potential, if it weren’t busy distracting people and lying to them with nonsense. I do not like this display at all, and I want them to fix it.

The display starts out well. It has a fair amount of complexity in it, conveying some several dimensions. I like this! It makes me want to trust a designer who put so much thought into it. Right off the bat I see a spatial representation of the river (the east to west alignment threw me for a sec, but doubtless it represents the system). I see creek crossings and creek tribs; if I knew the creek, those would orient me. There are boxes and they are sized to mean things, like the area of the watershed they drain. That is interesting and I can see it at once. So far, so good. I can see which tributaries are big and little, just by looking, so fast! I sure want to trust this diagram.

Except that triangle. I almost like the triangle, except… it doesn’t mean anything! HEY WAIT! That triangle! It doesn’t mean anything! It looks like it should mean something. Everything else means something. You know what it looks like, it looks like an accumulation of flow over the length of the river. That is the subconscious bias the triangle shape taps into. Who wouldn’t think of flow accumulation in a gaining reach? BUT! That is what the big arrow on the left means. The triangle is… redundant. Worse, it conveys an area. That area means what? Other shapes –the rectangles- mean land area in square miles. But the triangle doesn’t mean area of river bed. The shapes do not match; one is lying! The units of flow are AF-year, so the triangle means… the volume of water that passed through that reach? But it doesn’t look like that. It is smooth, no variance by reach. THAT TRIANGLE IS LYING! WHY IS IT THERE? Why is it there when the blue arrows are doing all the work?

Worse than a lying triangle was my realization that the triangle could have been interesting! What if the slope of the triangle sides varied by reach to show the accumulation of water or sediment or salts? Why, that would be so neat! You could see at a glance how reaches with lots of roads have steep increases in sediments or salts! The triangle would bulge and contract! The area in that reach would mean something, just like its friends the boxes mean something! Not the same thing, but we are sophisticated and can figure that out, maybe with a little help from some color or something.

And then! What if the slopes between reaches were actually from some function? All the display shows now is some measured data, and it lies to us about what that measured data actually does (with those long smoothed sides disguising trouble spots and implying that there must necessarily be an increase. LIAR!). It doesn’t have to be like that. It could be a function, like of swim-ability, which would be some combination of temperature and flow and pathogen content and maybe more or something. THEN! THEN, at every reach, little callout boxes could report each value in the function and the y-axis would be some measure of swim-ability at every cross-section.

HEY! What IF? What if you created a display like this, only you loved your stakeholders and wanted them to participate. THEN, since you have a function that creates the slopes of the lines between reaches, you could just ask them to changes the values in the call-out boxes! Tell us! How cold do you want your water in this reach (within a possible range, pending dam releasals and daytime temperature) and how much water and how much e.coli (depending on whether we crack down on those upstream trailer parks)? You tell us! In the model! And the triangle shape will change and you can play with it all you want. And if we had the infinite knowledge, we could tell you how much it will cost to make those changes in temperature and flow and pathogen content, and then you could see for yourself how much money you want to spend to get certain amounts of swim-ability!

BUT WAIT! Why is that triangle symmetrical about the x-axis? I mean, why? (Because the triangle shape fooled us again.) What does the height at a cross section mean? Twice the absolute value of some swim-ability index? Why? WHY NOT make the top function different! Like some fishing function of water temp, fish habitat, access to roads and whatever else! THEN, you would have different lines top and bottom, and users could change whatever variables they are interested in and we could know how much that would cost and it wouldn’t look anything like a triangle anymore! And the whole height would be an index of recreational use! God, it would be SO GREAT!

See? This display could be TOTALLY NEAT! It could draw people in and make them think about their creek and give them an intuitive response to the changes in values (you win the model when you push the top and bottom lines as far apart as you can for the money you have). Instead, that triangle makes people think less, for one, and deceives them, for two. I ordinarily like triangles fine, but I find that behavior unacceptable. I will reveal the truth about that triangle on The Internet; when information is free, people can judge for themselves. I think the people and history itself will judge that triangle VERY HARSHLY.


Blogger alison said...

Megan, I have no idea what you're going on about (though perhaps I could get halfway there if I made an effort), but I want to marry you even more today than usual!

1:13 PM  
Anonymous Dagon said...

Your anger may be misplaced. Geometry can't lie, numbers can't lie. They can be incorrect, or irrelevant (which seems to be the case here), but it's not their fault - they're just numbers and shapes!

People can USE diagrams and numbers to lie, by choosing meaningless or misleading ones. Sometimes this is through incompetence, and sometimes by design. If there's a consistency in type or direction of lie, I usually guess it's intentional. If it's just randomly pointless, incompetence is more likely.

So: who devised this chart, and what possible motives could they have for including incorrect and misleading shapes? Maybe they're TRYING to mislead you into thinking it's simpler than it is, or that you can't do anything about it.

Or maybe they just feel like it should be a smooth triangle rather than a lumpy intestine-like series of blobs.

1:16 PM  
Anonymous D said...

sad to say, the diagram in question feels like it was "adjusted" by a commitee... I believe the "Laguna de Santa Rosa Triangle" was probably added by the person who didn't think the original blue arrows conveyed the correct 'feel'. The person who put the blue arrows in in the first place didn't want to remove them for the sake of some 'weird triangle thing that doesn't mean anything'...

and so... to keep the parties from causing mass unrest in the office, their MGR approved it as is, because hey, he kinda liked the way the triangle made him feel about the whole slide.

so the takeaway is: do not make geometry lie to The Megan, for she is subtle and quick to anger...

2:00 PM  
Anonymous margie said...

What about the magic 25380 ac-ft that aren't shown on the right end of the figure? I just don't understand how you can be mad at a triangle, but couldn't care less that not all the numbers add up, esp. when over 10% of the flow contribution is not accounted for. What are they trying to hide?

2:23 PM  
Anonymous Anonymous said...

Since I'm out of town, I'm unavailable to whisper info design best practices in your ear, but you could do worse than curling up with Edward Tufte's books tonight.

2:52 PM  
Blogger Megan said...

Alison, that is my fault. I sorta blurted it out the way it occurred to me. If I'd taken the time to draw out each new concept I propose, an interested person could follow it better.


People don't kill people. Triangles kill people.


But they give us enough information to sort out the missing inflows. Sure, they skipped an arrow, but that isn't like making us think that the system is different than it is.

Incidently, y'all. Margie was adorable at lunch, when she picked up the copy of slide before we even talked about our weekends and studied it for several seconds, and then we had to talk about it before anything else.

2:53 PM  
Blogger Megan said...


One could always do worse than curling up with Tufte book.

2:57 PM  
Anonymous Ennis said...

This discussion was totally sexy.

3:15 PM  
Anonymous Anonymous said...

Why can't the big dumb arrow be used to just point in the direction of the river? What's wrong with that? It's a summary of the smaller blue arrows, and helps direct your eyes towards the flow of stuff.
-T. Chris

3:37 PM  
Blogger Megan said...

It is redundant with the smaller blue arrows. To display accumulated flow, it is fine. For salinity or sediment, it might be deceptive. Hold on, I'll put one up.

3:44 PM  
Anonymous doctorpat said...

This was possibly the best discussionb of a triange since that Pythagorus guy sorted them out some time back.

Looks like it needs sorting out again.

And I, for one, would thoroughly approve of a lumpy, intestine like curve that showed different data above and below the centre line

7:51 PM  
Blogger Tom said...

B-but... symmetry is good! It doesn't make my teeth itch! Unlike all those other, looser triangles. Eugh.

8:36 PM  
Blogger jens said...

For once, we have the same priors...

I have felt the same outrage at graphic arrangements that were just intended to LOOK GOOD and made you search for meaning in futility.

8:49 PM  
Blogger Megan said...

Unlike all those other, looser triangles.

I think the secret is that it shouldn't be a triangle at all. It could just be a couple lines running parallel to an abstracted creek.

10:39 PM  
Anonymous ed said...

Maybe it was just too much extra work to contour it appropriately. You know, those government have to leave them enough time to update their blog.

10:49 AM  
Blogger Megan said...

Well, it would require a function that no one has measured, estimated or derived. But you know, if you spent a lot of money to get that piece, the model could be REALLY NICE.

12:43 PM  

Post a Comment

<< Home