Creative Synthesis

We’ve mentioned Swivel and Many eyes in passing on this blog, but both of them deserve some further scrutiny. Sharing has become a big part of the web experience. When you look at the content being shared we see a very clear picture emerge. Photos, video, bookmarks, maps and more are all being passed back and forth, some more successfully than others. There are still some areas worth thinking about in terms of social sharing, and not all of them are as fun as photos. Some people might think data is boring. You wouldn’t know it from looking at Swivel and Many eyes. Both of these sites tackle data sharing with an eye toward visualization, expression, and communication. We’ve been sharing data on the web for a long time, although we’ve often encased it in web databases, textual commentary, and separate file downloads. These sites offer more direct tools for data presentation and sharing in an effort to build natural analysis not only through machine interpretation but through human communication. They’ve got people talking so let’s take a look and see what the big deal is.

Swivel: A place where curious people explore data

Swivel is a gutsy little site that is starting to get some traction. This is data sharing web-two style. Swivel has a strong focus on sharing and communication. Where the spreadsheet was designed as a ‘word processor for data’ swivel attempts to be ‘the web site for data’.

The good:

• Swivel gets sharing! Swivel manages to meet my definition of sharing. Given a situation where I have the kind of content they want (data and graphs) and it isn’t sensitive, there is no reason why I wouldn’t just host it on swivel and embed it into my page. It’s easy to do, worthwhile, and offers some nice visualization features if I need help visualizing the data.
• Swivel graphs are just images. This is both good and bad. On one hand, this means that sharing and embedding graphs is ridiculously easy and you can quickly drop them into other things (presentations, papers, etc.). On the other hand, they aren’t so interactive. The swivelian answer to multiple visualizations seems to be ‘build a new graph.’ This is ok, but services like Many eyes offer significantly more interactivity with their visualizations (but with significant limitations on sharability). All said I think I prefer the swivel approach, although I’d like to see an interactive offering that degrades down to their nice images.

The bad:

• Bad data and garbage graphs. There is a lot of data and a lot of graphs floating around the swivelverse. As of this writing there are over a million graphs to browse. This doesn’t have to be a problem but a lot of these graphs are variations on a theme, or come from differing versions of similar data. Better clustering features, more useful views, and better search options (worth a discussion of its own) could go a long way to making the community more useful.
• Some limitations on search. Swivel allows some browsing options (such as most viewed, most recent, this month and this weekend and by tag) but offers somewhat limited search options. With a natural language search you can limit what kind of information you want to look at (data sets, graphs, or users) but you have no real options to sort these results or perform advanced searches. This is a bit of a problem because not only does it make finding things harder, but it actually contributes to extra (possibly unnecessary) data. I ran into this problem in a relatively simple task. I was interested in seeing some graphs of data sets that had heard were recently released for second life. I wanted to search for ’second life’ data, but sort by the date posted. This was necessary because I wanted a copy of the data set that had been released two weeks ago, not the older data they had released. Currently such a search isn’t possible. After browsing through a few pages it was difficult to know which data set was the one I wanted, and I gave up (and put some of the data in myself).

  This is a somewhat specific example, but highlights larger problems with swivel. While I’m all about making lots of different views of data (graphs), the data set itself almost needs a little bit less sharing. I don’t want to go in and find 100 different copies of the same data set and worry about what the differences are. There are a few options that can address this:

  1. More metadata about data sets.
  2. Intelligent observation of entered data (”This data might already be available from user DataMonster, for example”).
  3. Better search and sort capabilities to prevent repetitive and unnecessary data duplicates.

The ugly:

• Swivel is big, not just in the number of data sets, but in the size of the elements. It’s hard to browse through and see what’s going on when we can only see six items above the fold. Admittedly this makes an attractive statement on the first use, but the lack of utility initially presented (there is a list view you can choose, although it is still not particularly svelte) suggests a focus on the new user impact, not the regular user usability.

Many eyes: For shared data visualization and discovery

Many eyes is a data sharing and visualization tool from IBM’s visual communication lab (which includes Media Lab alum Fernanda Viégas). Many eyes looks a little cleaner than swivel does. It is less overwhelming initially and the search results show more content for the screen real estate. This lets you get a better sense of what information is available. This is not to say there are better sort options, because there aren’t - but overall many eyes feels a little tighter and more focused.

The good:

• Dynamic! Where swivels graphs seem like static, almost separate entities, the visualizations in Many eyes actually feel like the direct result of the data set. Data is key in Many eyes, and even a simple browse of the data sets ties visualizations to the data and encourages the creation of new visualizations. I think that this is mostly the result of design and structure decisions. Swivel could easily do this, but viewing data sets on swivel by default produces the three in a row picture view. This is not nearly as useful or as informative as the Many eyes presentation.
  

• Visualizations: Many eyes goes a few steps beyond swivel in terms of visualizing data. In addition to traditional charts, Many eyes also offers treemaps, world or u.s. maps, block histograms, bubble charts and even network diagrams. Of these the treemap probably offers the most bang for the buck, and seeing it was quite a nice surprise. All of the visualizations are interactive. Not only does this allow more detailed analysis of data points, but in some cases it allows the visualization to be manipulated by large factors, showing the gross domestic product for the world counties and changing the year dynamically is one nice example.

The bad:

• Sorry java, I kicked caffeine. Well not really, I still drink significantly more coffee than I should but I’ve pretty much given up java on the web. IBM is still pushing java pretty heavily, and it shows here. The dynamic visualizations in Many eyes are the dreaded java applet. I can’t actually say they have bad performance (they don’t) but something about them just upsets me. Maybe I would feel differently in a world where Sun had supported java as aggressively as IBM does. I feel like the days of java on the web have passed and seeing these applets (even with there impressive interactive capabilities) makes the site feel dated
• It’s many eyes, not many sites. Sharing is becoming increasingly more important, but IBM doesn’t seem so interested in sharing. Don’t get me wrong. You can share visualizations from Many eyes. Unfortunately the sharing involves a small, nearly illegible, static thumbnail of the visualization that is linked back to the Many eyes site. This solution makes sense for IBM, they want everyone to visit the Many eyes site. It comes of as relatively useless and heavy-handed. I can’t say I would prefer sticking java applets on my site, but a more usefully sized visualization like the one swivel provides would be a nice touch.

The ugly:

• Did I mention the java applets?

Both of these sites could benefit from more sharing options. Swivel wins the prize here (and it is an important one) but I do like the ’snapshot’ thumbnails Many eyes offers up in theory. If they were a little more legible, something like sparklines, they could be a reasonable small scale link to the data or visualization. I think this is the takeaway message. Sharing the whole graph is useful, but it would be nice to have some more iconic sharing options for linking to data sets and sets of visualizations.

Both sites can benefit from adding more visualization options (in moderation). One of the nice things about Many eyes is that they obviously took some care in selecting visualizations that were useful. There is some overlap, but for the most part each visualization serves its purpose. There are lots of ways to look at data and adding more is nice, but only if they make sense.

I’m also somewhat surprised that flex doesn’t seem to be a player in the shared data site market. This is really surprising given the capabilities that flex offers for data binding and an impressive lineup of basic chart support. Even google has used some crazy flash visualization capabilities in the past for this information and I’m not sure if its absense says something about these sites or about adobe. These capabilities go far beyond just mouseover data point display. Quietly Scheming has a nice example of what you can do.

At the end of the day it seems that Swivel and Many eyes are positioned to serve serve different demographics. Many eyes comes across as more of a ’serious tool’ and a portal, where swivel is more about sharing and personal use, the portal component can almost be incidental. There is some significant overlap here, and I can see swivel moving more to being a better tool for visualization. At the same time, I don’t see many eyes becoming as supportive of sharing but that could happen too.

[Update: Robert Kosara offers a very detailed review from his perspective, which largely jives with mine.]