Assignment 4a: Evaluations

February 11, 2010

in Critiques,Student Posts

A Good example:

The periodic table.

Mendeleev's periodic table

Why do I think it’s a good visualization?

– It has been one of the most successful visualizations of all time. When it was first conceived, it exposed underlying trends which allowed scientists to design experiments and procedures to isolate most of the missing elements fairly rapidly, and it gave deep insights into the structure of atoms, allowing some startling predictions about undiscovered elements. The power of the periodic table is that it makes fundamental concepts so easy to grasp that it is used in middle schools to teach concepts that at one time were inaccessible to many of the most brilliant scientists in the world.

– It supports all kinds of queries very efficiently – as long as you know something about the properties of the element, you can go right to its entry, and get its atomic mass, electronegativity, symbol, name and other properties.

To use Munzner’s language, it works on several levels:

– It solves the right problem – the outer electron shell properties of each element are made explicit by their location, while the overall shape of the table shows patterns that govern all types of atoms. The proof is that many scientists were able to use the insights in the table to discover new elements.

– It uses the right abstraction – it is the same abstraction of electron shells that scientists came up with to describe electrons – that the number of valence electrons and filled shells can determine an element’s chemical properties.

– It uses the right visual encoding – position is first and foremost, and informative empty spaces are left in. Color can be used to encode other things, like the metal-nonmetal boundary, radioactivity, melting point, or other groupings.

– There isn’t really an algorithm to speak of…

Tufte’s principles:

– Comparisons – With the table you can compare elements, rows, columns, metals and non-metals, and the larger blocks corresponding to which subshells are outermost.

– Causality / mechanism – It’s got it in spades. Not to belabor the point, but the electron shell structure dominates most of the chemical and physical properties of elements, and it’s right there in the table.

– Multivariate analysis + Integration – Outer and inner shell electrons are encoded in position, and other properties are either shown in color, or listed directly in the table.

– Documentation – It all started with Mendeleev, but since then it’s been reworked by countless researchers, and is one of the most trusted and authoritative of all visualizations of data.

– Content – I think this principle is outside of the scope of visualization, but in any case, the content is a sound scientific theory, backed up by partial evidence in the beginning, and much more evidence since then.

A bad example

Statistical Challenges

Why do I think it’s a bad visualization?

– It’s a globe – you can only see one side at a time

– It looks like all nonsense – with other globes on the same site,, it at least makes sense to use a globe because they are showing data that has some geographic locality and “global” (not a pun) scope. Why would you draw these things on a globe? Where do they get these connections? What is the underlying story? What patterns are there in this?

– To be fair, this is probably not meant as a serious visualization, but as a kind of artwork. A lot of modern art seems to be trying to convey the feeling of helplessness in the face of the overwhelming complexity of modern life, but as a clear visualization of the author’s ideas, I think this one fails pretty badly.

Munzner’s principles:

– It’s not clear what problem they are solving, or if there is even an attempt to solve a problem here.

– The abstraction is a network between nebulous concepts involving various aspects of society and life in general. It’s only the least bit clear what the arrows mean – for instance, “standard of living” and “happiness” are only linked through the confounding influence of “energy production”.

– Encoding – position is not meaningful, color is not used, orientation varies without any apparent meaning, making it harder to read the text snippets, darkness (saturation) of the text varies in no discernibly meaningful way, and, being a globe, you can only see one side of it, and not without some distortion.

Tufte’s principles:

– Let’s just say Tufte would leave the room, and leave it at that.

{ 1 comment }

watkins February 11, 2010 at 1:44 am

This is a quote from the website you linked to:
“When [Mendelev] organized the table into horizontal rows, a pattern became apparent–but only if he left blanks in the table. If he did so, elements with similar chemical properties appeared at regular intervals–periodically–in vertical columns on the table. ”
I think this speaks volumes about the periodic table’s worth as a visualization. Not only did it organize what was already known about the elements, it correctly revealed parts of a pattern, and facilitated future discoveries to complete that pattern.
That being said, it’s usefulness is definitely restricted to people who already know something about physics and chemistry.

I agree that the globe must have never been intended for any serious reference. I mean, I don’t even know where the countries are in relation to the words. For that matter, it might not even be mapping geographic location. It might be trying to map causality, or some normalized amount of whatever is written… there’s really no way to tell.

Previous post:

Next post: