Gazing Deeply Into Your Many Eyes

Dear Many Eyes,

From the moment I stared into your thousands of solid black eyes, I knew we had something special. Since the day we met you’ve shown me the silver lining in my data and pointed out details that I never would have found on my own. You’re never pushy or arrogant about it; you always let me learn for myself. You believe in my natural pattern-finding ability the same way I believe in your big, beautiful exploratory tools.

Many Eyes, I want to tell you something. I just want to, well, let you know why you’re so high up on my bookmark list. You should also know there’s some ways that you can improve, but please don’t take it personally. I just want you to be all that you can be.

Sincerely,
Nathan

Most likely you’ve heard of Many Eyes, the online application by the IBM Visual Communication Lab, for shared visualization and discovery. Many Eyes offers a wide variety of useful visualization tools that let you explore your data in a very interactive way. This, Many Eyes’ focus, is why I like the application so much. At the same time, there are of course plenty of spots for improvement.

The Pros

User Responsibility

Data always have a story to tell us. The story might be something simple like an increasing trend, or the story might be complicated like multiple correlations to several variables. In either case, Many Eyes lets the data talk thru interactive visualizations. No assumptions are made about the data. Instead, the assumptions are left to the user.

The user also decides what type of visualization to use for her data. She has a choice among sixteen visualization options divided into six categories:

• See the world
• Track rises and falls over time
• Compare a set of values
• See relationships among data points
• See the parts of a whole
• Look for common words in a text

It often seems like more than sixteen options because with every option, the user can still explore her data from different angles (e.g. focus on different variables).

Social Data Analysis

As far as I know, social data analysis was coined by Martin Wattenberg, the IBM Visual Communication Lab group’s research manager. In a nutshell it’s “collective analysis of data supported by social interaction,” or in other words, it’s the idea of a bunch of people getting together (in person, via Internet, etc) around some data and then talking about the data. The hope is that even with everyone’s varying views, the group can come to some kind of intelligent consensus.

Many Eyes allows users to not only interact with data, but converse/argue/discuss about the data with others, and the visualization remains the center of attention. Two things about Many Eyes make this happen:

1. Topic hubs
2. Bookmarkable visualization

Topic Hubs

Topic hubs are essentially groups. Users can join a hub and then discuss and add data and visualizations to the hub. For example, I’m part of the “good data gone bad” hub. This hub focuses on misinterpreted or mishandled data. This allows people with a common goal or interest (e.g. economics, politics) to get together and give a go at some data.

Bookmarkable Visualization

The second factor, bookmarkable visualizations, I think is the most useful and important part that contributes to the socialization of data. A difficulty with interactive visualization is that asynchronous collaboration can be difficult. Say you’re exploring your data and you find something interesting. If you later want to show that to someone else, you don’t want to have to tell them to click this, then that, then go there, and then, uh, oh wait, go back, now forward… On Many Eyes, you can take a snapshot. Someone can click on that bookmark, shooting her straight to what you were looking at.

The Cons

Data Reliability

As with other applications of its ilk, Many Eyes still has some work to do on data reliability. However, I get the impression (please correct me if I’m wrong) that Many Eyes isn’t trying to be a data archive. They mostly focus on visualization. I mean they just recently put in a data editing feature, which seems like it would be a top priority for an application trying to be a data warehouse. No?

If I were a new user though, it’s possible I could be coming to Many Eyes only for data since data sets are tagged and listed right under visualizations. They can also be saved as tab-delimited plain text files. Interactive viz relieves some of this burden as it allows users to see possible kinks in the data, but surely there is more that can be done in the background to assure legitimate data sets.

Data Uploading

Along the same lines, data upload needs improvement. I like Many Eyes’ instant feedback feature when I cut and paste a properly formatted tab-delimited text file to the text field. Cut and paste is fine (sort of) for small data sets, but for large data sets, it could be problematic. There’s a lot of things that could and will go wrong as the data makes its way from a spreadsheet to the browser. A user might accidentally not copy all of the data, some points might get entered incorrectly, etc. Hence, I think users should be able to upload their data as files. An acceptance of other data formats would be useful as well but probably not as urgent.

Minor Gripes

The user interface is still a bit clunky and the look and feel could improve. The links, which seem to make up a lot of the text, is a bright orange that can be hard to read. Perhaps darker, bolder colors might work better. Browsing could also use some work. Visualization and data sets are only shown as a bunch of thumbnails or lists, respectively. Surely, a visualization application can come up with something better. Something that displays by popularity? Maybe a mosaic?

Anyways, now I’m just nitpicking. If you’re still reading, I thank you for making it this far down the post. I’ll stop for now.

Many Eyes is an excellent application, but I think it can and should aim to be more than a collection of exploratory tools. I’m looking forward to seeing Many Eyes develop, and if you haven’t explored Many Eyes yet, I encourage you to do so.