Information Density and Dr. Bronner

Edward Tufte, in his new book, Beautiful Evidence, continues on his crusade for information density. Here's a representative recap of a Tufte seminar from 2001:

This is a companion discussion topic for the original blog entry at:

Bob wrote:

“Yeah, but birth control? Is that really SO necessary while hiking?”

Rest assured that I agree the whole Dr. Bronner package ranting it totally whacked. I’m not a fan of castile soap anyway, so I don’t like what’s inside either. That said, I’ve heard that hook-ups are not that uncommon on the Appalachian Trail. :slight_smile:


There seems to be something akin to “lean” thinking here, aiming for maximum information flow, but there is still the constraint of usability, as per Donald Norman’s ideas on design. Too much information puts people off, results in mistakes, etc. Also highly dense displays are unlikely to be accessible to visually impaired people or those with low-res displays (mobile phones, web browsing
by speech/audio, etc). I don’t think density is the key. I think matching the display to people’s cognitive processing is. Pie charts beat tables for proportion comparisons. Graphs are better for trends.

I would agree that displaying as much information on the screen is a good way to go if you don’t have too much conflicting and different information to convey.
As with any debate, things tend to go to the extreme, each side using the extreme of the other to argue their point. “Your website has a million and one things and is confusing to navigate. Thus we should go minimalistic”. When in reality, it only tends to be confusing if the million and one things are unrelated, and the interface is trying to do everything at once.

I think there is a difference between a busy interface and a dense interface. I think the latter implies that it is filled with relevant, useful information. While the former implies a lot of noise (graphics, ads, icons, etc.) To me, Yahoo is busy.

But I think there is something missing from the discussion: sure, a page in a phonebook has a high density of relevant information, but its a very specific purpose for a very specific type of media. You certainly wouldn’t expect to port a phonebook to the web as just a series of pages to click through. The web interface would use search and show relevant results, albeit with less density than a printed page.

Comparing computer screens to printed media is only applicable (IMO) when the computer representation is static (Word, PDF, scan, image). When you are talking about an interactive application, the rules of the game are different.

I think Steve Krug of /Don’t Make Me Think/ fame answered this question for me: it doesn’t matter how much is on the main page, so long as users can find what they are looking for quickly. I believe he even cites some usability testing he did where they saw that users only really scan and thus only really see the thing they are looking for.

Usability has to be the king here. If users can’t access the information we’re trying to provide – no matter the desnsity of the offered data – then it’s like the data is not even there.

A phone book page can hold 36K of information, but how much of that is USEFUL to the reader? I’d say about 7-100 bytes (depends if you’re looking for a phone number or an address). So we’ve got this book full of probably billions of bytes of information, only 100 of which are useful to the reader at a given moment. Compare that to a search engine where I can type the name of the store (or the type of store), and scroll through maybe 5000 bytes for the 100 I need. The phone book is progress compared to that?

Tufte is the classic example of somebody’s who too smart by a half (i. e., he’s smart, but not smart enough). What he fails to grasp is that not everybody is like him. Where he see “Beautiful Evidence” many other see confusion. They simply can’t deal with the data density he finds so illuminating.

With whom are you trying to communicate? Yale academics? Then go for the density. Or are you trying to reach out to people who can’t keep multiple mental balls in the air at one time? Then go for simple and direct.

A phone book has a dense data representation out of necessity. You simply have to put as much information on a page as possible because you are limited by the physical format of the book. The page has a fixed size, and a total number of pages that is fesible to use.

How do you search a book? You quickly fan the pages till you get to the section with the letter you want. Then you flip pages till you get to theright one. Than you put your finger on the page, and scroll down till you find the information you want.

Compare this to Google interface where you get the information in two steps. You type in a query, scroll down the results list.

The information density on a web page might be lower, but the speed at which the information arrives is much higher.

There is no need to put allot of information on a search page because most people only stay on that page as long as it takes them to type in a query.

Is it desirable to have your user fishing for information on your page? For example, I routinely use Firefox’s incremental search feature to find the details I want when I encounter a page that overwhelms me. What I’m doing is taking your high density page, and extracting information I want.

Sometimes this might be a desirable setup. If you are presenting statistical data, or some kind of compilation, it might be logical to put it all in one place and let the users fish.

But I definitely do not want to use my browsers search feature to find the news section on your page or hunt for the illusive download link.

I’d have to say Tufte’s just wrong. I’ve seen several of these text-only-cram-it-in sites, and they make it harder to find stuff. Like everything, there’s a balance, but subtle visual cues and good use of negative space get the user to their data quicker every time. The problem is these solutions get ‘bottled’ so quickly, then re-used without thinking about their application.

Digg is a good example- tons of content, but enough space and color to get an idea of what’s going on instantly. But witness the crop of copy-cat sites trying to apply the concept and failing, either because implementation is poor, or because the whole concept doesn’t apply in their situation.

Well i think there is a fine balance between content and the interface to that content.

It all starts with knowing who your end user is.

Know them, be friend them, and learn how to continually improve how you deliver what they want, that you have.

If your a organic soap buyer, so obviously their design/interface is not gonna be aimed at you.

But yet design can’t be too unusal, that it detracts new customers from using it.

So it’s a balance. But so much of this feel of balance is just instinct, intuition, it’s hard to put factual basics into it.

I believe in informational architecture, how the page is structured, easy to read, easy to find what i need to find or want to find.

If i can’t find it, easy access to help or support.

The only reason to make it difficult is if that is part of the allure, like orkut, hidden or hard to get access to.

That can be fun at the start, but in the end if the users can’t figure out how to get hte products/services they want in an easy enough way, your competitor will easily snap them up.

This really seems like the “age-old” fight between print designers and web designers.

A phonebook is a static medium where the only interaction between the “user” and the medium is the flipping of pages and the reading of text.
The phonebook entries are sorted and categorized so it is easy to find the information your are looking for. A phonebook is easy because it’s entire contents are (usually) in one large volume, right in front of you.

Thumbing through a phonebook is a LOT faster than clicking, clicking, clicking, clicking through multiple pages of information on a web site. Can you imagine if you had a database of a phone book and then just dumped it’s entire contents to someone’s browser, forcing the user to scroll, scroll, scroll page after page, trying to duplicate the functionality of your thumb on a book?

When you have an interactive site, where information is contained in a database or on some other site, you need to have a more sophisticated way of searching for and retrieving data.

So Tufte doesn’t like navigation bars, but wants choices. Choices of what? Navigational aids are nice because:

  1. If a user is “scanning” text on a page, they will know pretty quickly if what they are looking at is what they want.

  2. Why force them to continue to read or scan to look for navigation to take them somewhere else?

  3. If you have well thought-out and executed navigational aids, a user can use that navigation to help find their information much quicker.

I think a lot of people are missing the fact that Jeff is basically musing out loud and not advocating it; all the comments are preaching to the choir of anyone who would read this.

Shawn: The Do Not Call list does exactly that, opens up a text file with every number on the list in an area code. . (They eventually added a searchable interface, but never bothered to point it out.)


Yeah, I know. But I am sometime like the Dennis Miller of development and I go off on rants sometimes.

I just dont “get” people that don’t “get” that the web/your monitor is a different medium than a newspaper or a book. Trying to force a web site to read like printed material is like the whole “square peg in a round hole” idea.

Keep making the square peg small enough, and it will eventually fit through, but it’s not exactly satisfying.

David Ogilvy (sp??) of Ogilvy and Mather, advertisers extraordinare, published widely on this 50 years ago for a couple of decades. He concluded, and research at the time confirmed, that the more you told the viewer the more successful the ad was. Same idea.

We reactionaries argue that pixel glitz doesn’t add any kind of density, save noise.

The main point about use: either support all input from the mouse, or not at all. It’s the flipping back and forth that makes people crazy.

I’ve always wondered whether the amount of text on the Dr. Bronner’s soap package is based on an awareness that the product is very popular with backpackers/hikers, particularly on the Appalachian Trail in the U.S.

Hikers use this stuff for everything–soap, tooth paste, dish washing, deoderant–because they only have to carry one product. (I tried brushing my teeth with it once and just about lost my hiker’s breakfast.) Hikers have lots of time to sit around reading.

It also reminds me of the religious rantings you see in “folk art” such as that from Howard Finster.


Yeah, but birth control? Is that really SO necessary while hiking?

Over the time ,the people are giving up to the evident: the data density will must match the data we can process. Leave the upper density to the machines, they can handle it, but they cannot understand it. At last, we are talking about better comunication from interface to people…

Minimalist content certainly sucks, and it is definitely a capstone of the web 2.0 movement. Sometimes however people mistake the amount of content for the density of content.
Just cause someone is using 400 words, doesn’t mean they are providing any information. I’d much rather read “Upload your photos here, free, for life”, than …

“We are a web based image application processing firm from Ohio. Using our compementary combination of bleeding edge web technologies, you can upload any graphical data you see fit, in compressed or uncompressed form, and we will endeavour to ensure that your data remains intact for a duration that we see fit.”

It’s basically about saying everything that must be said, as succinctly as possible.

I attended Tufte’s course in Washington D.C. last year. I found it to be an excellent course in presentating information and overview of his printed material. It was a truly enlightening experience for me, a web developer.

He specifically addressed digital displays and their inability to match print in terms of resolution. He asserted that digital displays are a long way off from matching print resolution. However, he noted that digital displays typically allow for interaction. He suggested providing summary data that can be summarized and then drilled into or zoomed into in order to reveal more precision. Also, he suggested that non-web interfaces (mission control, etc.) require multiple display units (see ). The best of both world’s would be interactive digital display with print-like resolution or print with interaction, but until those options exist, we need to make design decisions appropriately.

A lot of people like to bash Tufte without reading his material or attending his courses. They don’t realize that the majority of the examples he cites are printed and span the course of history. It’s critical to really understand his material before critiquing it.