Unless you live under a rock you’ve probably heard that Congress voted down Hank Paulson’s bailout plan and then the Dow lost 777 points. The fundamental question is: why was it voted down? There’s a lot of speculation that it’s due to the deal being unfair, it being too kind to Wall Street, etc., but I think there’s a more subtle issue.
I think the bailout failed in part due to a failure of metaphor. No one in the Administration came up with a nice metaphor to succinctly explain to the average American why they needed to buy assets they don’t understand to keep a tightly coupled complex system afloat.
The one I keep thinking of is the gas station analogy. The banks are the gas stations and credit is gas. If you run out of gas or it gets really expensive you’ve still got your car and it works but you can’t get anywhere or you end up taking the bus. Maybe this is too politically sensitive a metaphor in these days of $100 gasoline, but I think Main Street could be pursuaded to with a good metaphor (or simile).
A little over a week ago I attended Edward Tufte‘s day-long course on how to visualize information. What follows is a chronological summary of what he presented; the course consists of a series of infographics along with running commentary.
The first infographic was actually an animation: the Music Animation Machine‘s rendering of Chopin’s Berceuse, opus 57 (a lullaby). You can’t see that video online, but here’s a rendering of Chopin’s Etude, opus 10 #7:
There are two powerful elements to these visualizations. The first is that you cannot delete anything from the graphic without removing information. The second is that you can literally choose to hear a note by looking at it.
The next graphic was the following one from Nature (apparently the best source to see people pushing the limits on infographics) showing the epidemiology of SARS:
What makes this graphic so useful is that it includes labels on the linking lines so that it subtly adds additional information. The use of directional arrows also clearly indicates the nature of relationships. It’s not perfect though – it could add incident and mortality rates in different countries and not overwhelm the rest of the graphic.
A quick commentary: if you want to create illegible boxes, give them a big, heavy border and use all caps sans serif fonts – that what cigarette companies do on their warnings:
After that we looked at a graphical history of rock & roll:
There’s a massive amount of information on this chart – more than any one person can possibly explain. What makes the chart work is that every reader has a chance to look at it and explore it using their own cognitive style. Each user will tease out their own story; if they can’t find one then you can tell them one.
Another related point: clutter is not a property of information; it’s a property of design. If you’ve a cluttered diagram you don’t understand your information.
The next graphic was a hospital bill overlaid with commentary as to what each element actually meant (alas, no image online). The bill was a two column itemized list; the commentary was provided in boxes next to the components of the list.
Tufte emphasized that the connections between the boxes were via gray – not black lines; this rendered them more visible. The point here was that a grey line was the smallest “visual move” you could make to connect to the graphic and not distract from it.
From there we moved on to a graphic showing cancer survival rates (here’s a sample):
Note that the graphic goes from least to most lethal and visually depicts how a person can expect to survive over time. While this may seem trivial, you won’t find a chart like this on any major health-related website. And, in fact, Tufte mentioned that he used to show up as the number 3 result in Google queries for “cancer surival rates” – well above any health-related sites.
And here’s Tufte’s favourite graphic of all time. Charles Minard’s description of Napoleon’s disastrous advance on Moscow:
What makes this graphic so impressive is that it has six variables on it: location (x&y), time, the size of Napoleon’s army, temperature and direction. Without a word, the horror of Napoleon’s failed march is immediately obvious. Note that there’s no flourish on this map: it’s just data.
From there, Tufte lectured on a variety of topics. Here are some random comments:
We do our best analytical thinking when we are 24-30 inches away from the page. That should help you decide how big a graphic should be
For serious work, you need data that is adjacent in space (think of a wallchart), not stacked (like a typical Powerpoint presentation)
Whenever possible, get a ‘relevant object’ in the room when you’re giving a presentation: we are human and love both tactile objects and metaphor
Sidenotes are “where god wants footnotes”
When designing a user interface, always focus on what percentage of total screen real estate is data/information (vs. marketing speak)
Genius of the iPhone is that it reduces the need for stacked paths. Most phones had the same functionality but made it impossibly difficult to get to. One of the reasons why Apple was able to do this is that the screen has 2.5x the resolution of a computer screen (hence the screen is effectively 2.5X larger than a similar-sized phone)
If possible, don’t present rather give people a report ideally containing a “supergraphic” like the Rock & Roll history map. They can read faster than you can speak and will ask questions if they find anything interesting in the supergraphic
See how long you can go in your presentation without saying “I”. Will make you appear more informed and professional
Use the “PGP” rule: say something particular, explain the general principle and then say something particular. This is how teachers educate young kids and it works well
The course also featured a number of great quotes. Here are two:
John Von Neumann: There’s no sense in being precise when you don’t even know what you’re talking about.
T.S. Eliot: Talent imitates, genius steals. Apparently Eliot stole this from Oscar Wilde: Good writers borrow. Great writers steal.
And now for some not-so-serious info. Check out this little message that was left in one of my books over lunch:
I love bands who dance to their own music. It’s as if they wrote their own songs as nobody else could create anything they could listen to. Over the years I’ve found that these bands tend to put on the best shows.
And so it was last night at Webster Hall where Cut Copy tore the place apart (on a Monday night!). The overwhelming feeling was that they loved their own music – but not in a narcissistic way. Rather, it was as though the band had a goal to make people dance and their music just happens to be the best way yet they’ve found to get a thousand people to dance together.
Every day I walk past Rozzo & Sons fish wholesalers. It’s a furnace of nasty smells on Ninth Ave. By the time I’m going to work they’ve been up for hours and the street has been reduced to a sea of boxes, melting ice and fishy water.
There are two things I love about it. The first is that it harkens back to when this was a working city and the sidewalk was an extension of your shop-and possibly your workshop.
The other thing that I love is that it reminds me of how small the world is. Check out the photo below. A day or two ago, a boat in the Falklands came back to port with a fish. Hours later it was on a plane to New York. Shortly after it was sold in the Bronx at the new Fulton Fish Market and then it ended up passing through a store on Ninth on its way to a restaurant. The only trace is a box left lying on a Thursday morning.
I’ve recently been playing around with the Django web framework (all non-techies should stop reading this at once). Here’s a little snippet of code that’s useful if you want to add a timestamp to a model you’re creating:
from datetime import datetime
class My_Model(models.Model):
date_created = models.DateTimeField()
date_modified = models.DateTimeField()
def save(self):
if self.date_created == None:
self.date_created = datetime.now()
self.date_modified = datetime.now()
super(My_Model, self).save()
What happens is that when you save the object, you override the Django save method with your own. It checks to see if the object has a date_created timestamp. If it doesn’t (i.e., you’re creating it for the first time), it adds one. It also updates the date_modified timestamp every time you save it.
You might be wondering: why not just set the model to look like this:
It won’t work due to a quirk in Python. Python will calculate “now” to be the time when the model is loaded into interpreted. This means that every single object, until you restart the server, will get exactly the same timestamp. Not too useful – so use this hack instead.
So I went away for the weekend and someone in my neighbourhood was murdered. However, before you panic, you can rest as the suspect is in custody. In fact, he’s being interviewed right now. This is the scene outside Precinct 10:
You can’t tell from this photo, but there are in fact six tv vans plus another one circling the block. It’s an absolute zoo…
No, this post is not going to have anything to do with girls. Rather, it’s about Gregg Gillis a.k.a. Girl Talk. His latest album, Feed The Animals, is a sampling assault. Each song jumps from sample to sample at an amazingly frantic rate; it’s been on my iPod for weeks now and I can’t get enough. It’s music for people with ADD (if you like it, I recommend Arthur King‘s Pop Sh!t).
The challenge with his music – and any sample-based music in general – is trying to explain it to anyone who’s never heard anything like it. Fortunately, the good folks at Wired magazine are up for the challenge. Check out this visualization of one of his tracks; you can find it in the latest issue (16.09):
Finally saw EyesWideShut only 12+ years after it's release. My confusion over its meaning led me to this awesome screed http://t.co/xF0e9u0r42 years ago
RSS
Recent Comments