Temporal Interactions I: History

From CS260Wiki
Jump to: navigation, search

Lecture Slides

File:Cs260-slides-17-time1.pdf

Extra Materials

Grossman, T., Matejka, J., and Fitzmaurice, G. 2010. Chronicle: capture, exploration, and playback of document workflow histories. In Proceedings of the 23nd Annual ACM Symposium on User interface Software and Technology (New York, New York, USA, October 03 - 06, 2010). UIST '10. ACM, New York, NY, 143-152. DOI= http://doi.acm.org/10.1145/1866029.1866054

Nakamura, T. and Igarashi, T. 2008. An application-independent system for visualizing user operation history. In Proceedings of the 21st Annual ACM Symposium on User interface Software and Technology (Monterey, CA, USA, October 19 - 22, 2008). UIST '08. ACM, New York, NY, 23-32. DOI= http://doi.acm.org/10.1145/1449715.1449721

Akers, D., Simpson, M., Jeffries, R., and Winograd, T. 2009. Undo and erase events as indicators of usability problems. In Proceedings of the 27th international Conference on Human Factors in Computing Systems (Boston, MA, USA, April 04 - 09, 2009). CHI '09. ACM, New York, NY, 659-668. DOI= http://doi.acm.org/10.1145/1518701.1518804


David Kurlander: Editable Graphical Histories (Video)

IBM History Flow Project

Discussant's Slides and Materials

File:Discussion Section Temporal Interactions I.pptx.pdf

Reading Responses

Airi Lampinen - 10/29/2010 13:34:07

Hill's and Hollan's "Edit Wear and Read Wear" article presents two applications that are build around the notion of "computational wear". While physical objects get worn out in use, often in ways that are intuitively informative, such wear is not an inherent characteristic of digital goods. The authors discuss ways to introduce computational wear to applications, in order to allow for a "reflective conversation".

In brief, Edit Wear shows where a digital object, be it a text document or a spreadsheet, has been most heavily edited, while Read Wear creates traces of which parts of a document have been read most. These features are intended to make it easier and more intuitive to see what the current "hotbeds" of working on a certain digital object are. This is deemed to be potentially useful both for individual and especially collaborative use where it might otherwise be hard to see what others have been up to. The idea of introducing automated traces of editing and reading is appealing, even if for some purposes intentionally and actively created notes are sure to work better.

Thinking of collaborative editing, such as writing a conference paper with multiple co-authors, it would in many cases not be enough to see which parts of the texts the others have worked as it is essential to understand what the underlying thoughts and goals have been. On the other hand, quick ways of seeing what has happened in a document since one has last been editing it is useful - think, for instance, about Word's Track Changes feature that allows for reviewing changes without needing to reviewing the entire text.

As to reading, the benefits are perhaps more questionable because the mere time spent on reading a certain section might mean very different things. The section could, for instance, be the least clear one in the text and hence require re-reading, or, it might be the best and most interesting section and hence attract the readers attention for a longer while than less insightful bits.

Adar et al's "Zoetrope: Interacting with the Ephemeral Web" describes Zoetrope, a system that enables interaction with the historical Web. The basic idea of the system is to provide a temporal search to the web, looking to changes through three types of lenses: visual, structural and textual. The paper discusses in some depth the challenges related to searching the web as pages change continuously in a number of ways and explains the authors' solution to overcome them.

The concept of Zoetrope is interesting and I don't doubt that a search like this would be useful for a number of purposes. However, I was puzzled by the complete lack of a discussion related to privacy as regards to the effects Zoetrope could have. Increased information access is in many ways a positive development but not an unproblematic one. The capacity to search the historical web as outlined in Zoetrope might have huge implications, so I would have wished for the authors to discuss their thoughts on these.

Also, the examples in the paper are very strictly oriented towards "informative" (or do I dare say neutral) information, such as traffic and sport scores. Yet, we use the web increasingly for social purposes, too. This is a further reason why ignoring privacy issues and other social implications feels like a big flaw in this paper.


Pablo Paredes - 10/31/2010 13:26:56

Summary for Hill, W. and Hollan, J. - Edit Wear and Read Wear

This paper describes a system to incorporate the notion of information physics in a document, which will reveal the effects of interactions over time in documents. The types of information wear implemented are edit and read wear. These notions are also presented in light of Schoen's professional work theory of reflective conversations, which analyzes the way interfaces support problem definition versus problem solutions. Schoen argues that by revealing information tool usage new tacit discourses can be revealed, which will support the possibility to reformulate hypothesis that will improve problem definition, rather than limiting the interaction to work on implied problem settings to focus solely in finding solutions.

The authors discuss certain attributes of these type of implementation, such as the notions of ownership/choice and simplicity/efficiency. They define the need for clear notion of ownership by the person's role, i.e. a reader has access to read wear attributes and an editor to its corresponding ones. They make a point that weariness should be embedded in simple and efficient ways, such that the notion of weariness doe not affect the use of the tool, but rather enhances its communicative power. Finally, they propose the need to rather embed these attributes in existing tools rather than new ones, in order to reduce dependency

The overall CSCW theses presented by authors, where Edit Wear and Read Wear could improve interaction rather than imposing social processes remains a big curiosity to me. The argumentation in favor of revealing tacit information in order to improve dialogue and empower problem definition seem powerful and natural to me. However it seems this idea illustrated almost 20 years ago has not raised to a level of either implementability or adherence, and it is not clear to me why... The only notions of mass readability I have seen is lately with tag clouds, which I indeed find useful and revealing.

I think that new tools that record and reveal timely and historical interaction data could generate very interesting and new ways to work on collaborative environment, or even help a single user observe his own tendencies and inclinations to do better introspection of his/her work. These types of tools could also be very beneficial for future work of social support groups, where patterns of usage could be traced to reveal groups with similar characteristics that could later be united via some types of traditional or innovative tools to generate communities of practice, therapy groups, creative think tanks or any other types of interest-centered groups.


Summary for Adar, E., Dontcheva, M., Fogarty, J. and Weld, D. - Zoetrope: Interacting with the Ephemeral Web

The paper describes the implementation of a system that allows a user to capture patterns of information built from historical time-series of information. The work describes a systemic approach with a language semantics and operators to facilitate different queries. The authors define the types of information capture as lenses and the operators as filters. They define lenses that could work in a geometrical space (visual), in a html segment (structural) or in specific information tags transformed in variables (textual). The filters could use factors of time, content, volume, pruning duplicate content, logic relationships among the previous ones and triggered filters working in cascade. The grammatical constructions are built graphically and in a very simple fashion, using boxes, highlights and lines to generate many valuable elements of information.

The system implementation consisted of a crawler attached to HTML/DOM files and images, which generated a content stream filtered by operators and rendered on user interface windows. implementation of the system demanded certain careful considerations, specially those related to information retrieval, information display and storage. Information retrieval demanded extraction schemes that could perform rapid crawling. The system coupled its crawling intensity with time based on an analysis of the survival rate of DOM elements. In the case of intermittent changes in pages (temporary promotions, alerts, etc), information retrieval must be able to fill in the gaps in such a way that information display does not generate a notion of incompleteness, an artificial solution based on filling the gaps with content that is closer in time was chosen by the authors to give a notion of completeness, however a small indicator (a red tag) revealed this technique to avoid the notion of deception. Regarding information display, one of the biggest decisions was how to position smaller lenses in terms of its relative position with the complete website. Instead of moving the website content, which would have generated distraction by prompting the user to check the complete screen, rather than focusing on the lens, they decided to create long time sequences similar to historical lines, which allow to see the change over time with no background distractions, but which could be also shown as little movies to reveal time-sequential patterns. Finally, regarding storage, a key question was the need to define some sort of "information delta" which captures only the marginal changes of websites in order to avoid large storage needs.

Overall, I find this work remarkable. I think this type of analysis of the web contents has tremendous potential to reveal many human knowledge dynamics, which could have impact in many human activities, ranging from business intelligence to science. Additionally, the simple and well thought interface design reveals the careful and sensitive approach to incorporate interface design as part of the system itself. As a matter of fact, it is wonderful to see how simple graphical elements, such as rectangles and lines have tremendous power and clear grammatical sense to create complex queries with very little effort. I personally would really like to use this system to incorporate it in my research about human behavior in ecosystems such as families with similar problematics. Seeing the type of trends in online information and compare this to the types of behaviors in real life could be of great interest. I am still concerned that the enablement of the systems and the information storage may prove to be the biggest entry barrier to these types of systems, therefore backend infrastructure business models may need to be tightly coupled with this service.


Krishna - 10/31/2010 16:17:51

Edit Wear and Read Wear

The authors describe a user interface system that shows edit and read histories of a document using attribute mapped scroll bars. The basic idea is to show edit and read histories as marks on the scroll bar. While showing edit wear, the length of the marks are relative to the number of edits and for read wear, the length of the marks depict the amount of time spent by users reading portions of the document parallel to the marks. Furthermore, the system allows grouping edit and read wear information into categories.

Their implementation of the system was straightforward. Edit histories were recorded by recording every edit of a line and indexing it by timestamps and categories. Read histories were recorded by tracking, roughly, the amount of time a line was visible in the editor's window. This implies that all the lines visible in the window get relatively equal scores. This does not sound intuitive - they could have weighted reading time scores based on where a line is in relation to the current position of the scroll bar. It is not clear how the system automatically assigns categories to the wear information, I am assuming this is determined using information of the current user reading or editing the file.

The authors claim that their interface aids reflective conversation and problem-setting, emphasized in Schoen's Theory of Professional Activity. In other words, they argue that patterns of read and edit activity provide feedback and help users 'adjust what they are doing'. They fit their idea into a computational wear framework, inspired by physical wear, and suggest that wear information, which come free of cost without extra effort, can act as useful metaphors in user interfaces. They also discuss interesting generalizations of the idea by applying it to menus and spreadsheets. For example, read wear on menus can provide category selection statistics and act as recommendation metaphors. Finally, they discuss problems that may arise from recording extensive histories of edits and reads - problems such as information overload, ownership and privacy issues.

Zoetrope

The authors describe Zoetrope a system that allows exploration of web pages over time. The interface shows webpages as images in a zoomable canvas. Past versions of the page can be seen by using the sliders at the bottom of the page or by using lenses over specific areas of the page. Their system architecture indexes webpages and timestamps them; pages are stored as both DOM structures and images. They have a suit of operators and renderers which filter, transform and provide visualizations while displaying the image within the interface. Their architecture is extendable and developers can implement their own operators and renderers and plug them into their system.

The most novel part of their interface are the lenses. Users can create a lens by drawing a rectangular region on the image. By using the sliders users can get a history view of the selected region. The lenses can operate both at a region level - track changes to that particular region of the webpage or operate at a structural level - track changes to a particular DOM element. Both these operations assume a certain level of stability over structures or regions of the page. To track unstable regions, the system offers a textual lens - lens that track arbitrary text regardless of where or how it appears in the page. Interestingly, filters can be applied on top the lenses - for example, users can track content that is distinct from content seen earlier, etc. Furthermore, lenses can be bound and stacked together across multiple pages - an interesting use case is that users can track traffic at specific time during specific events by binding lenses over sites that show traffic and sites that show sporting events.Another interesting feature are the visualizations - page histories can be viewed using timelines or even animated as movies and the clustering visualization allows using information from one lens to group information in another - for example, group traffic patterns over weather patters.

An interesting, creative work. However, I am not sure about the computational requirements of the system - though they have described such issues to a reasonable extent, the scalability of the system for widespread use still remains. Also, semantics of interactions between multiple pages using only the DOM is not very clear. But it is understandable that their focus was towards designing an extensible, pluggable UI framework that facilitates interacting with temporal samples of multiple web pages. Great read.


Thejo Kote - 10/31/2010 17:20:01

Edit wear and read wear:

In this paper, Hill and co-authors present the concept of "wear" in digital documents. They refer to wear in its physical sense, i.e, indication of use. Edit wear represents the history of changes made to the document and as the name suggests, read wear refers to the readership history of a document. The authors propose a graphical way of displaying this information through scroll bars. Since sroll bars already represent the relative length and position in a document, they overlay information about edit or read wear in the corresponding section on the scroll bar. They call them "attribute mapped scroll bars".

They also provide theoretical perspectives of the work with regard to Schoen's theory of professional work, an information physics view of interface design and its illustration of a CSCW thesis. Being unfamiliar with the former two perspectives, I'm in no position to comment on them, but with regard to CSCW, they propose that using techniques they present may do away with a need for some social processes which enable co-operative work. Knowing who is editing or reading certain parts of a shared document is certainly very useful in a collaborative setting.

The idea of annotation of scroll bars is a pretty popular UI metaphor in modern systems. We see examples of this in the indication of the occurence of search terms in a document, indication of warnings and errors in IDEs etc. But, there isn't a temporal aspect to the use. I can certainly see the benefit of the concept of "wear". Even more interesting is the "wear" of menu item usage in complex applications like word procesors or spreadsheets. There are definitely interesting ways to improve the UI that one can think of in those applications.

Zoetrope:

In this paper, Adar and co-authors present a system which enables interaction with historical versions of a web page. Users now access the latest available version of a web page. The goal of Zoetrope is to enable temporal interactions with a web page. It allows the creation of lenses on specific parts of a page, and to view the changes to that portion over time. This has many use cases like finding the changes in a numerical values over time, viewing co-relation of data in the same page or across pages etc.

Zoetrope allows users to filter results in order to make the results more relevant. It also provides visualizations of the data in time series and other forms. The authors describe their implementation which uses a DOM and image based system.

This is a very innovative system. The use cases they describe are very compelling. At the same time, I think there are challenges to usability in the way they present it. Apart from the basic creation of a lens and application of simple filters, I think the rest of the system is not for the novice user and maybe it wasn't designed with that intent in the first place. There are also clear scalability issues for a system built for the web, but I think that issue can be adddressed. The bigger issue, I thought, was the general applicability of the solution in the wild. As they mention in the paper, web pages change all the time. This system is likely to be useful for short periods of time only. But, overall, I enjoyed their approach to temporal information retrieval.


Aditi Muralidharan - 10/31/2010 17:46:13

Temporal Interactions I: History

In "Edit Wear and Read Wear" Hill et. al. present the interesting idea of allowing electronic documents to reflect their usage patterns - i.e. patterns of reading and editing, by means of a sidebar with a bar plot, or heavier or lighter coloring on spreadsheet cells. This idea is so interesting that I wonder why modern word-processing programs don't give us the ability to see this, and modern web pages don't have the ability to record how people read them.

This paper felt very relevant to me as I was reading it. It struck me how useful it would be to see marks of my own read wear on academic papers I had read for project when I was stuck, or writing up a paper, or trying to remember how a particular problem was approached. In the more general problem of supporting the information seeking process, showing the user their own trails through material they've read could be useful.

In this class, previous discussions seemed to conclude that Vannevar Bush's idea of information trails was adequatately implemented by hyperlinks - but I don't think that simply linking documents together was what he had in mind. Read wear could be a a way to show a reader his or her own progress through a forest of information - much like a hiking trail through a real forest.

The second paper offers a different take on tracking changes in document content over time. Zoetrope is a system that allows users to see through "lenses" on a section of a web page to see previous versions of that page or track content over time. I like the idea of being able to refer to sections of a web page just by dragging over them. Like the visual search engine/scripting utility Sikiuli we discussed last week, it a step towards making the meaningful contents of GUI's or rendered pages directly manipulable. I really like the idea of being able to select-and-drag a section of a page and have it not just mean the surface contents of that section, but the contents, in a "deep" sense, of those page elements.

One clear limitation of tracking history by page structure is what happens when the structure of a page changes - if markup standards change, or is not consistent across different parts of the site, it would break.


Charlie Hsu - 10/31/2010 17:47:09

Edit Wear and Read Wear

This paper describes the concepts of "Read Wear" and "Edit Wear", metadata tracked by a prototype document editor that provides information on who has read/edited a certain line of a document. The paper's implementation displays this document information in the scrollbar of the document editor. The paper then explores some of the theoretical implications of "Edit Wear" and "Read Wear", such as their impacts on Schoen's theories on professional work, how "electronic wear" can emulate the information provided by wear in physical world objects, and how they illustrate that shared artifacts are better than process control for small group collaboration.

The comparison between physical and computational wear was the most interesting point in the paper for me. The authors brought up a critical point about physical wear; since it is a by-product of normal activity, it essentially has zero cost to create. Though every operation on a computer theoretically has a cost, we can consider many of them negligible in the temporal domain; keeping track of the wear in storage is, as the paper predicted, a cost that is shrinking dramatically with cheapening storage. In the physical world, developing products that keep useful wear data may take a lot of effort (one example that comes to mind are shaving razors with physical indicator strips that wear down with use, sending the user a message to change his blade), but in the digital realm, creating metadata is far simpler to do and only requires frameworks and the ingenuity to realize some data is worth keeping (i.e., the read/edit data of a document).

Document editors and source management control systems today all keep track of "wear" data that may benefit the user. Google Docs keeps track of authors, edits, and concurrent readers. Source control systems keep detailed logs of edit history, document versions, and user comments. However, as hinted in the paper, one of the remaining problems to be addressed, even today, are the continual iterative design improvements that could be made in choosing which data to store and how to display it. The paper mentioned the possibilities of using color and texture mapping rather than simple scroll bar drawing. Today, we have all sorts of document metadata constantly tracked, and storage is cheap enough to save it all. What sort of data, and interesting visualizations of said data, can best inspire the "reflective conversation" the paper cites from Schoen?


Zoetrope

This paper presented Zoetrope, a system that allows users to interact temporally with the Internet. Zoetrope uses an hourly crawler on several popular websites to offer temporal interaction with the Web. Users can then set lenses, filters, and visualizations to draw relevant information out of a set of temporal Web data. Examples are shown displaying the many different tasks Zoetrope's small set of operators are capable of.

I found that Zoetrope's main contributions to the HCI community were the interesting operators chosen to navigate temporal data in a Web perspective. Using the DOM tree structure of HTML and lenses of visual, structural, and textual types were all strong exploitations of the Web's structure. Clustering and examining time series were both great applications that took easily parsable media from the web (numbers, static images) and created interesting statistical visualizations from otherwise relatively inaccessible information on the dynamic web.

I feel one of the key limitations with Zoetrope, however, is that the ephemeral Web still limits the amount of temporal data one can gather. Users or developers of Zoetrope must specifically define which websites to track (tracking them all is certainly not possible), and it does not seem practically possible to be able to retrieve temporal data from a website had Zoetrope not ben previously crawling it. Zoetrope's system seems to be relatively storage-efficient (100kb per crawl). One possible solution might be to offer some sort of central database or peer-to-peer interface for sharing temporal web data. For example, if I recently discover a news site that many others had known about before, I could use their cached temporal data to immediately receive the benefits of Zoetrope's data visualization and filtering.


Dan Lynch - 10/31/2010 17:55:36

EDIT WEAR AND READ WEAR

This paper introduces a very interesting perspective when it comes to documents---the wear that a reader or author leaves on a digital document. A very interesting approach using a scroll bar to indicate the amount of wear in different portions of a document. Additionally, collaborative aspects were discussed in how this type of visualization can benefit co-authorship.

I think one of the best applications for this would be the co-reader use. For example, when reading a conference paper, you could see where the most readers have spent their time, and in which paragraphs. This could help you get a great deal of information fast.

ZOETROPE: INTERACTING WITH THE EPHEMERAL WEB

This paper introduces a system that allows people to explore the time dimension of a web page.

For many practical uses of this system, I think this problem is intractable. For example, many queries may require cookies or authenticated logins. Additionally, get and post requests are often used to generate web content, and the set of all combinations of these in infinite.

You shouldn't have to index web history. That information is stale for a reason. Most likely information that has changed did so because the preceding information was outdated or incorrect. However, I will give the idea credit where its due. For historical purposes of major web pages, like the CNN web site, does have its benefits. But in that sense you can use the internet archive which has been indexing web pages since at least the 90s.


Siamak Faridani - 10/31/2010 17:56:18

These two papers are an example of building systems on the basis that was developed through many years of research. The first paper, Eidt Wear and Read Wear, connects former research and theories in the CSCW systems to define and refine its theory about Edit Wear and Read Wear. The second paper, which is published 16 years later, takes the idea and extends it hyperlink systems to build a new system based on the attribute-mapped scroll bars.

Attribute-mapped scroll bars are similar to the idea of sparklines by Edward Tufte. It seems that authors have patented their attribute-mapped scroll bars and I am wondering if using similar ideas in the second paper caused any patent infringements. I am also wondering if Tufte’s sparklines infringed authors’ patent rights. I believe presenting data about the history of an attribute in a very condensed area is much older than Either Tufte or Wroblewski’s papers.

Both papers are closely related to Schoen’s theory, They allow the reader/author to see the evolution of the article and follow the direction and the path that the author originally used. As a result the user is exposed to the pattern of editing and reading activities as the document takes shape.

The concept of lenses and filters that are introduced in the second article seems much richer than just simple attribute-mapped scroll bars. Authors extend lenses to many dimensions, visual lenses, structural lenses, and textual lenses. This allows them to mix and match and combine different lenses and it enables them to use different operators on these lenses (similar to relational algebra :)

We can somehow see the same implementation in today’s wikis, although they do not use any graphical tool to visualize the evolution of articles in time, they use a textual timeline to give editors and readers the background about the state of the document that they need. Readers and editors can see the history and take the document to the next step. I am surprised though, if attribute-mapped scroll bars have such a rich background why systems like wikipedia do not use them. I also found the references to ethical issues very interesting. These read and edit wears capture the whole history of a document, as a result any errors will be visible in the history of the document for a long time.


Bryan Trinh - 10/31/2010 18:15:01

Edit Wear and Read Wear

This paper presents a technique of displaying historical document data that follows Schoen's theory on professional work.

When first reading the brief about Schoen's theory of professional work, I immediately thought about the papers that we read on embodied cognition and prototyping. This idea that having a shared artifact to collaboratively work with is stronger than other indirect ways of communicating changes--speak through the documents in this case. I very much like this idea, and it seems strange that nothing like this has been implemented successfully in a document editor. This system also does not require any extra work from the users. Perhaps in practice, it is hard to see the benefits and effects of this extra data.

I think that is the complication though, how can the developer display this extra data without being distracting? The scroll bar idea seems like a very good solution, but the spreadsheet mock-up could easily conflict with other data that is mapped to colors. In that regard, it seems as though the major problem is displaying this extra history data in a way that the users can use effectively.

Zoetrope: Interacting with the Ephemeral Web

Zoetrope looks like a phenominal tool that anyone can use to monitor and track temporal data over time. I was stunned to read about the time lenses that enable the user to scroll through and plot data that is changing over time. With this tool, anyone interested in any sort of numbered data on the internet can quickly analyze data. I can't wait to see the production version of Zoetrope.

I am floored at how awesome this thing is right now, so I can't think of any blind spots.


Richard Shin - 10/31/2010 18:25:49

Edit Wear and Read Wear

This paper discusses the concept of 'computational wear', or the visual depiction of the history of interactions with, and manipulations to, documents. Specifically, the authors present 'edit wear' and 'read wear', which aim to present to where others most read and edited a textual document to the user. To achieve this, they use 'attribute-mapped scroll bars', which overlay markers of proportional size on the scroll bar to indicate the magnitude or frequency of activity. The paper further discusses how the authors implemented the system in a variant of Emacs, and its theoretical implications relating to professional activity, physical wear, and CSCW.

I realized that I actually used this idea of putting markers in the scroll bar on a daily basis—Google Chrome offers the same functionality for in-page text searches, marking points in the page where the query can be found. At its core, this paper's main contribution is the idea of marking scroll bars, and that certainly seems like a useful one suitable for displaying any form of data that varies along a document. The implementation of the system itself described in the paper didn't seem as interesting in comparison, though, and the theoretical implications that they consider largely concern how this system fits into existing theory, rather than proposing something particularly new.

Unfortunately, it didn't seem that the authors had performed any user studies on how people use and benefit from their system. The benefits certainly would have been clearer if we could concretely see how users performed editing or reading tasks more efficiently than others (although, I suppose it would have been hard to demonstrate, since accurate data is essential for this system to be beneficial). Even on an intuitive level, I'm unconvinced that read and edit wear would help with text reading or manipulation tasks. With read wear, for example, it seems that the wear marks could become self-reinforcing; if the marks are shared, for example, people might scroll to where the markers are only because there are markers there, and not because the content is particularly useful or interesting.

Zoetrope: Interacting with the Ephemeral Web

This paper describes Zoetrope, a system that enables temporal navigation of web pages. Zoetrope allows viewing the history of web pages, either entire ones at once or through smaller 'lenses' which show only a particular visual region of a page, a part of its document structure, or some arbitrary piece of text no matter where it may have moved on the page. Zoetrope also includes various forms of filtering for the lenses, such as on time, on keyword, whether another filter was matched, and so on. These tools enable temporally-aware visualizations of web pages, such as headlines shown on a news web site, a graph of a counter increasing over time, or traffic patterns given certain weather conditions.

The idea of viewing old versions of web pages is not new, but the interface mechanisms introduced in the paper certainly seemed that way. Since many parts of a frequently-changing web site remain static, such as the logos and navigational elements, the lens enables focus on whatever is of relevance. The 'scent' markings on the slider, conceptually very similar to the marked scrollbars in the previous paper, allows the user to easily navigate to relevant points in time, while showing the temporal relationship between them.

I'm not convinced, though, about the general utility of a 'lens' as a metaphor for viewing into the past. It specializes in monitoring how a certain visual or structural slice of a web page changes, but doesn't provide all-encompassing information about what changed on the page. With the example of maps given in the paper, examining a small part of the page on its own gives all the information needed. But with a blog, for instance, it doesn't seem very useful to see which post had appeared in a certain part of the article listing. While the visualizations discussed in the paper would better address this concern, they have the problem of being too removed from the web page itself to be directly useful there. Also, it seemed that past versions of web pages were simply stored as images, discarding many of the interactive aspects of web pages; the authors briefly discuss how hyperlinks work with the system, but it was unclear to me whether interaction with the lenses was possible.


Matthew Chan - 10/31/2010 18:30:46

===Zoetrope: Interacting with the Ephemeral Web===

This paper offers a novel way of interacting with the world wide web. This has a new perspective where users engage not only the current html pages, but also past versions. This potential is very promising, because the authors highlight a very important point that users can't look back in time, such as previous traffic condition or baseball rankings.

The techniques used was a web crawler for 250 websites every hour, and the interaction technique was drawing a window. In some ways, this is comparable to a camera where a user can see a snapshot into the past. Briefly, the ability for a user to draw a frame was a design decision chosen over loading the entire page because the former addresses events when the user uses several frames. Moreover, the authors created three interaction techniques such as textual, structural, and visual lenses. Other features enabled are clustered and timeline visualization.

Perhaps another fun feature is to extrapolate into the future. Although more entertainment, it would offer a place for artificial intelligence to try to predict the future. On another note, this paper reminds me of something Google did on one of its birthdays: the offered users a link to the old Google page from the early 2000 with the same search results from that year. My friends and i were playing with it, such as searching for "MySpace," which didn't exist yet. However, Zoetrope offers a better interactive technique with the window frames and binding them. This can also change the way how search is conducted since we might be able to look into the past.

This paper does not relate to my work in any way. The only blind spots i see are possible limitations since we would require lots of storage to keep track of the world wide web over the last couple of years.


Edit Wear and Read Wear

This paper was a very entertaining read because it strongly relates to work i did over the summer on a project called HyperSource which helps developers document their source code after visiting web pages. This acts as a live documentation by associating webpages to lines in the text editor (we used Processing instead of ZMacs).

The paper is pretty important because, similar to the metaphor of a "wear," it offers other users and developers to understand the critical parts of a document. However, it's strange that since 1992, we still don't see anything like this in commercial products like Word or Office. The benefits are great if we could see the hot spots or stable spots of other readers, but i see a possible security/privacy issue. This is similar to the events right after September 11, when the intelligence agency or the FBI wanted to gain the rights to track suspect's receipts or documents from libraries or book stores to track what they read. I consider this to be a potential blind spot that the paper doesn't cover because the Edit/Read wear might track too much (a similar issue that HyperSource had in its earlier implementation).

In a similar note, Edit/Read wear made use of the scroll bar to visualize hotspots on the document, which was very similar to what HyperSource: we made use of the left side bar to denote which lines were associated with a URL or command line prompt.

Overall, i think it's fascinating that 18 years later, a similar concept (HyperSource) has built on an idea almost 18 years ago. It does baffle me as to why Read/Edit wear didn't take off. More specifically, the paper did not run any user studies to check the validity or effectiveness of Edit/Read wear and only considered theoretical perspectives and Schoen's work.


Luke Segars - 10/31/2010 18:35:13

Edit Wear and Read Wear

This paper discusses two simple but powerful ideas about using user behavior to improve a particular expreience, described in this case as editing a text document but also applicable to a wide variety of other processes. The two ideas, called Edit Wear and Read Wear, involve monitoring user(s) behavior during a particular task and using the accumulated information to provide information to new users. In particular, the authors mention that it could be used to mark relevant or dynamic places with a simple visual signal, metaphorically equivalent to the smudged pages in a physical book.

This approach seems particularly important for applications dealing with large amount of data because it allows the analysis work of others to provide added value to new users. Major sites on the internet today such as Google and Amazon have tremendously useful implementations of user wear that allows users to reduce the number of possiblities that they need to search before finding something valuable. In a way, this technique is crowdsourcing an investigation of the importance, value, or relevance to the users themselves so that it can be provided back to future users. This paper was published around the time of the emergence of the internet to the masses, the perfect time for a technique that required a large number of actors on single documents. The arrival of the large-scale internet drastically increased the frequency that users viewed and analyzed the same content by providing static web sites that could be viewed by anyone.

It will be interesting to see how far the scope of this technique can be extended. Many of the tasks that we perform today are both online and collaborative, suggesting that this technique could be extraordinarily useful in other cases. For example, could certain icons or widgets be resized or emphasized (or irrelevant ones hidden) if a particular group of unconnected users finds them to be particularly important? Could we do more with recommendation systems using this approach? It's exciting to read about and see how this idea has already materialized, and it will be equally exciting to see how much further we can make it go.


Shaon Barman - 10/31/2010 18:39:09

Zoetrope: Interacting with the Ephemeral Web

Zoetrope provides a way for users to track and query past versions of a website.

The basic concept in Zoetrope is the lens. A lens provides a view of a certain part of the html field. Once a lens has been specified, queries can be made which filters out versions of that specific lens. This allows the user to quickly retrieve information from past versions of a page. Lenses can also be connected so that queries are made on multiple lenses.

One major difficulty which arises when finding semantic meaning in the web is that the structure of the html for a web change can change dramatically without the actual image being altered significantly. Javascript also adds to this burdern. This makes it difficult to relate the html to what the actual user sees. Zoetrope overcomes this obstacle by allowing the user multiple ways of selected in the lens, many of which can be customized to overcome html changes.

A few obstacles I see with this approach is the difficulty in specifying filters and lenses and the lack of transparency in what is being done. Most browser users know little about HTML and the DOM. Because of this, it would be difficult for users to know when each type of lens would be useful and to fix broken lenses. Also, if such a technique did become popular, it would restrict web designers from making massive redesigns because such designs would break this tool.

While Zoetrope does enable users to create queries that would have been difficult or impossible to formulate in the past, it seems like most of these operations should be implemented in a domain specific manner. If there is an actual need for a certain temporal feature to be recorded, the site owner could cache the original data and allow the end user to access it in a way similar to Zoetrope. This would allow all the benefits of Zoetrope, but with a more intuitive way to specific lenses.

Edit Wear and Read Wear

This paper proposes a visualization which shows how often certain lines of text in a document are revised or viewed.

The visualization seem very intuitive, by embedding it into the scroll bar they put more information in the UI without any extra cost in screen space. The idea of explicitly added the use of a digital object seems appealing, and could be added to all sorts of computer interfaces (such as a music playlist or file browser). It is almost a way of sorting objects, but it is done automatically in a way which is obvious to the user. Another dimension of use which might be worth adding is visualizing how recently a certain section has been edited. I can imagine scenarios in which a heavily edited document can be misleading if all the changed in the past month occur only in one section.

It seems like the scroll bar is an overlooked UI element which has much potential. Because of its purpose, it provides an abstract view of the document. On this abstracted view, more information can be added. This was done for edit wear but other metrics could also be implemented into it. It would have been interesting to see f visualizing the edit and read wear affected user performance in any way. On one hand, its nice to see what has been worked on but sometimes this extra information can be bothersome and actually hinder performance. Overall, this paper provides a innovative way of presented information and asks questions about whether the information that is usually disregarded (such as edit wear) can be beneficial to the user.


Anand Kulkarni - 10/31/2010 18:43:47

Zoetrope


Zoetrope is a system for visualizing, exploring, and querying the historical state of web content.


Zoetrope offers a lens interface for interacting with previous versions of a webpage, for instance, letting users stream historical versions of the DOM, which is terrific. I think the novelty of this interface is itself a useful contribution, since it adds a deeper, temporal dimension and a relatively direct metaphor for interacting with various historical states of a web page. I like the contribution of using "recomposable" logical operators for making queries on historical states of web pages -- this strikes me a novel and a natural way to study the evolution of web pages. I wish that the authors had better emphasized the novelty of their crawler; for individual users, being able to explore their past web history is highly useful.

Zoetrope's authors present a substantial discussion of its feature set rather than an extensive evaluation from an experimental standpoint. As a proof-of-concept, it seems to be given as an existence proof. I like that the authors incorporated several modalities into its operation, which provide stronger support that the system is both novel and useful. At the same time, it would have been more valuable to have an experimental validation of some kind. For instance, I would have liked to see a validation of a given user's ability to trawl historical data from their own past week of web activity (for example), or to see some extended use scenarios.


Edit wear and read wear


The authors discuss new tools for visualizing the amount that different portions of a document/menu are edited or read on an ongoing basis.

Edit wear has a natural, fantastic application in tracking the evolution of Wikipedia pages. Read wear also has a major application in evaluating the usability of webpages. These contributions could have a major impact on the way we interact with documents in the future, if, for instance, they were incorporated into web browsers and word processors. I like the potential for these systems to improve the performance of writers and content creators as they go about their work, and I think they could have a direct impact on the way people carry out these tasks; I wish the authors had explored these possibilities further.


There is no evaluation per se, but rather an exploration of the theoretical implications of this work in terms of Schoen's theory of professional work and how the two tools developed demonstrate that group collaboration is better organized by shared artifact. This strikes me as an inappropriate evaluation strategy for this pair of tools; given that they have obvious applications in terms of interfaces or improving writing or usability testing, they should have been explored quantitatively from these perspectives. The analogy of computational wear is certainly an interesting perspective and valid under the theories the authors present, but are still rooted in an inappropriate evaluation metric.


Brandon Liu - 10/31/2010 18:46:13

Edit Wear / Read Wear

The most interesting part of the Computational Wear system is how there is no action needed on the part of the user/author to add data. This gets around some of the major problems with CSCW systems where the burden of adding information is placed on those who do not benefit from that information. Systems like this are already ubiquitous in programming environments like Subclipse, but are only starting to emerge for general documents.

One issue with the system is how it can infer the ‘read’ state from the user. The implementation described is not fine grained, since it only works at at the granularity of the displayed page. It is difficult to imagine how this could be improved without gaze tracking.

Overall, I like the idea of making virtual documents more like physical artifacts. One could take the ‘physical artifact’ metaphor and go farther. The system only encodes the amount of reading or editing relative to the others - it doesn’t encode absolute time. For example, a physical document shoved into a back drawer may accumulate dust - this notifies the reader that the document has not been touched for a long time.

The paper discusses privacy issues at a surface level. In the case of aggregate statistics, the data is anonymized and there is no attribution of views/edits to individual users. Systems like Google Docs already show who is viewing a document at a certain time, as well as who was the most recent editor.

Finally, I would have liked to see more discussion of evaluation of the system. It would have been nice to get user responses showing that the more relevant parts of a document could be more easily found.

Zoetrope

Zoetrope adds or replicates the ‘history’ functionality a lot of sites already offer. For example, Amazon does not show price history on items. Stock prices usually have histories associated with them. In the case of item price history, I can see how Zoetrope would be useful. In general though, a history system made by the application developer would better fit the data.

A potential use for Zoetrope is cross-website analysis of temporal patterns. For example, the number of a certain t-shirt sold on Amazon, around the time of a popular internet meme. This kind of analysis would go beyond what individual website developers could add to their applications, and best use the affordances of Zoetrope’s data.

Many websites already encode temporal information - for example, a blog like Gizmodo has older stories down the page, and a ‘more stories...’ link that takes them to even older stories. An interesting facet that Zoetrope explores is not only the timing of changes on a website relative to each other, but also the rate of those changes. This is implied in the ‘Scented Widget’ that shows new matches as tall bars and matches as short bars. The ‘Timeline’ features of Zoetrope seem like they would need a lot of consideration in their design - specifically, the need to encode time as absolute or log-scale.


Arpad Kovacs - 10/31/2010 18:57:54

The Edit and Read Wear paper describes a system that records how users interact with a document, and then on future uses show the areas in which past interactions occurred. The primary technique for doing this is attribute-mapped scroll bars, which show marks corresponding to "wear" on the scroll bar, with the width of the marks corresponding to the magnitude of historical edits in that location. This helps users identify stable (rarely changing) and hot/controversial (frequently edited) areas of documents, as well as which people last read the article, and when.

I thought that putting marks in the scrollbar was a clever and effcient reuse of existing space. As the article mentions, this could be quite useful for cscw applications, where providing the user with feedback on which sections of a document his/her collaborators deal with could be very useful. However, there are some questions that the article does not answers. How does the system show deletions and moves? Would vandalism (eg deletion of the entire article, and subsequent reinstatement) make this approach too noisy for high-profile applications such as Wikipedia? How is it possible to see how the article changed over time, rather than just aggregating all of the edits up to a particular point in time? How can we see what parts of the document a single user interacted with?


Zerotype introduces the concept of lenses that allow users to interact with previous versions of a page through direct manipulation techniques. The user first creates the lens by drawing a rectangle around the area of interest, and then he/she can specify how far back in time to look using a slider. The system allows filtering by time, keyword, amounts, as well as allowing duplicate elimination, compound filters, and triggers.

I think that from a technical standpoint, this paper is very impressive. By caching a combination of xml and image data, Zerotype can quickly and efficiently search textual and image data from the DOM, while presenting the results to users in a real-time visualization interface. This seems very well suited to websites where the user can predict the that changes, such as train schedules, weather forecasting, and stock tickers. However, I doubt that it would be too useful for standard story-driven websites, which have too much variance for the lens to provide meaningful information (eg it would be hard to use Zerotype to extract useful information about how the front page of the New York Times changes over time, since the stories are so different).

Overall I think that this is a clever idea, however the applicability seems to be limited. Also, for this service to be very useful, it would have to make frequent updates of a very large array of websites, which as the WayBack machine has shown, is quite a challenging task.

However, I am not so sure this would work for pages that


Luke Segars - 10/31/2010 18:59:10

Zoetrope

This paper describes a system that can be used for collecting temporal "states" of the web as they change over time. This application, called Zoetrope, crawls websites at an established interval and stores the data so that it can later be processed, analyzed, and filtered. Although there are some interesting potential applications (mostly scientific), there are some obvious problems with it that may make it inaccessible to the common user.

Perhaps the most useful application of Zoetrope that I forsee is the ability to analyze web data over time as an relatively inexperienced web user. Several of the analysis ideas mentioned, such as tracking product prices or traffic conditions, could be both useful and interesting to a significant number of people. You can imagine how visualizations of this sort of data could influence commuting and shopping behavior simply by making the data available, and the extensibility that Zoetrope promises suggests that here are many more applications that could be created by curious users. It would have been interesting to have the authors explore the potential applications of this technology more and discuss whether they could see it heading in the future outside of some of the more obvious uses.

Nevertheless, the same problem exists (probably moreso) for interested individuals as exists for the site that hosts the data originally: it takes a *lot* of storage to hold onto the state of even a small slice of the internet. As time passes, the space requirements continue to increase. While it may be possible for large corporations (Amazon, etc) to store some of this data, I suspect that it will be difficult for users to collect meaningful amounts of information from multiple sites without running into some problems. This may be mitigated as the cost of storage continues to drop, although the pace of information exchange over the internet is likely to increase significantly as well.


Thomas Schluchter - 10/31/2010 19:01:48

Zoetrope

The authors present a system that allows end-users to track changes in content and structure of web pages over time.

Overlaying the web with a timeline is an interesting concept. The researchers managed to make this usable enough that end-users could conduct "time-series analyses" of web pages. On a larger scale, this is remarkable because it means that the ordinary end-user web can be used as data, something that the Semantic Web has been promising for a while. It mostly failed because the barriers were to high: homogenous rules for authorship and shared understanding of the structures of meaning that are used to relate data from different sources to each other.

Of course, if the Semantic Web had come to fruition, the relatively resource-intensive approach of Zoetrope could have been avoided. It seems fairly unlikely that this could scale to a large population simply because of the demands on storage if every state of any web page were to be cached for later analysis.

Edit Wear and Read Wear

The paper discusses a system integrated into a word processing application with which uses of a document are recorded and visualized in context. In analogy to objects of the physical world, said use is inscribed into the documents.

I found the underlying idea fascinating. The usability of collaborative software such as wikis suffers from the fact that the change going on in them is recorded separately, as an abstraction from the artifact in question. Giving people a sense of which sections of a document have experienced the most change as they interact with it is like embodiment on the virtual level. Knowledge about the document is accumulated through its use and alters its appearance at the same time.

I wonder why the researchers confined themselves to scroll bars for text documents. The argument that screen space is precious is definitely true, but one could imagine breaking the information down to finer levels, and to directly embed it in the text itself. For highly contested Wikipedia topics for example, it would be interesting to see even changes on a sentence or word level to visualize the ongoing negotiation of meaning. If this renders the document illegible, adding the information as a layer that can be activated as needed, seems like a reasonable solution.

Drew Fisher - 10/31/2010 19:02:31

Zoetrope: Interacting with the Ephemeral Web

I see this paper as taking the idea behind the Internet Archive (archive.org) and making it more easily accessible for quick data analyses. It adds quick mockup functionality

I liked the idea of being able to track changes over time, but the amount of infrastructure required to accomplish this task seems painful. Further, one has to have crawled all this content in the past, if one wishes to do any data analyses. This limits the scope of this research to either

  1. datasets provided by parties that already run large-scale webcrawlers, or
  2. researchers who have very specific things that they want to index.

These practical considerations severely limit the broader impact of this paper.

Another thing that will break this system is the likely event that a website will change its layout at some point (and as time progresses, this grows more and more likely). Another item that will break screenshotting is anything that has animated content (although these things are less likely to change in interesting ways over time).


Edit Wear and Read Wear

While knowing what has changed recently or often seems most useful to someone who is actively editing a document, or someone who has read it before. This might prove useful in software documentation, but for general communication, I don't see that much value in edit wear.

Read wear is complicated by the fact that reading a page does not necessarily imply that the reading done proved useful. That said, it could help optimize first readings of a subject, directing you to what others found useful. I see this as being rendered in our world today as search engines analyzing clickthrough to adjust search results - you aren't aware of it, but you're using that read-wear information that has been collected. On the downside, these are self-reinforcing systems - the most-read portions of a text would be those indicating that they should be read most.

I'm not sure that making Edit Wear and Read Wear obvious to an end user are valuable so much as the concept of observing the changes of a document over time, and using that knowledge to provide intelligent interface choices.


Matthew Can - 10/31/2010 19:03:50

Edit Wear and Read Wear

This paper presents two document processing applications, Edit Wear and Read Wear, that visualize the history of edit and read operations performed on a document. These applications are an example of the concept of computational wear.

As the authors describe, these applications are motivated by Schoen’s theory of professional work. This theory emphasizes a reflective conversation with the problem in the context of the problem-setting. Edit Wear and Read Wear create a richer problem-setting because they make it easy for the user to identify patterns of edit and read interaction. This, in turn, enhances the quality of the reflective conversation.

In addition, this paper’s work is also inspired by the useful information that physical wear leaves on real objects. Augmenting digital artifacts with computational wear can allow for the creation of useful information as a side effect of normal use. But, one useful property of digital artifacts is that they do not have to experience wear, unlike physical objects, which cannot avoid wear. We need a better understanding of the benefits and drawbacks to computational wear.

In particular, I think this paper would have benefited from a user study. The authors talk about the potential benefits of their applications in an abstract sense. What I would like to know is how they actually help people read and write documents (perhaps they are merely distracting), and how they compare to after-the-fact statistics logs.

On the topic of CSCW, I think that computational wear can possibly enhance workflow (again, a study is in order). I agree with the authors that computational wear has the benefit over process control tools that it does not restrict the work process and does not require continual updating. But, I don’t see computational wear as a replacement for process control tools, but rather as a complement to them.


Zoetrope: Interacting with the Ephemeral Web

This paper presents Zoetrope, a system for browsing the historical Web. With it, users can explore the history of a web page, or use a temporal lens to view the history of some part of the page. Users can apply filters to lenses and bind multiple lenses together. The system also provides a means to visualize the history of a content stream.

Although I’m not sure if this sort of system is widely applicable, I do think temporal web browsing is an interesting concept. However, it also has many challenges, so I liked that the authors described some of the ways that their system deals with the changing structure and content of web pages. In particular, I am referring to the structural and textual lenses that Zoetrope supports.

Something I gleaned from the paper’s example usage scenarios is that this system can be used to support a kind of end-user programming on web pages. For example, instead of writing a script to alert when stock a stock has reached a certain price, one could simply create a lens and apply a filter. Of course, this would require a higher sampling rate for pages like Google Finance.


Kenzan boo - 10/31/2010 19:22:36

Edit wear and read wear.

This article proposed a way to track changes and view them. essentially tracking how the author edited an article and presenting it to the user, the reader, in a useful manner. it proposed a way to alter the way the scroll bar looked to provide more information on when the changes were made in time and where the changes were made so the user can then track it.

this was written in 92, by now there have been several implementations in writing software that try to track this. even using things like version tracking systems does the this function, but in a less readable way. searching through a page in mac's preview does a similar scroll bar indication of where the targets are too. there is a lot of room to discover what is usable in this field to track time without overwhelming the users with too much information.

Zoetrope: interacting with the ephemeral web.

this article was about tracking the web through time and archiving either all or specific target sites to retain information of how the web has changed. one example of this that they suggested was the need to see what the traffic conditions were like in the past at a certain time of day to get a gauge of what it will be like in the future. they proposed a system that a structural indication of the page, and once focused on an element, it would provide detailed temporal information. they could also parse information like numbers to generate a useful graph of the changes.

the biggest challenge to this will be archiving everything and having the space to do it. once that is done, providing an easy understandable interface to traverse back in time also becomes challenging given the huge amount of information there. however their solution seems pretty well thought out. the key element of this would be search and filtering. without these, there is no way to possibly understand and go through such a vast amount of data.


David Wong - 10/31/2010 19:39:53

1) The "Edit Wear and Read Wear" paper describes a method of keeping track of editing and reading activity for a document. The paper describes the implementation and the theoretical grounds of the system. The "Zoetrope" paper discusses the Zoetrope system, a way of temporally querying the states of web pages. The paper discusses the features of the system, its implementation, and related work.

2) The "Edit Wear and Read Wear" paper brings up an interesting point, however, I don't know how useful this is to the HCI community. The paper was written in '92 and since then, we haven't really seen any of the innovations they described in document processing software. In theory, the idea of keeping track of edits and reading is valid, however, in practice it is another story. I suspect that it didn't provide the added-value theorized and as a result, never found it's way in to production, commercial software. Nevertheless, the idea is interesting and can possibly be usefully applied in HCI. The "Zoetrope" paper is interesting theoretically, but I don't know how big of a contribution it is to HCI. The idea of temporally indexing web pages is not novel. Their suite of querying, filtering, and visualization tools, however, offer a new perspective on the issue. However, I don't think that their system could scale if placed in practice. Also, the validity of it's results is subject to when and how they crawl. Altogether, it's a cool tool, but I don't know how useful it really would be.

3) The "Edit Wear and Read Wear" paper argued the theoretical side of their point. The paper illustrated how their system coincided with the Shoern's theory, but never gave any counter examples. As such, the argument was clearly illustrated, but I wasn't fully convinced. The paper did not describe how their system would actually add-value in practice, which makes their argument much weaker. The "Zoetrope" paper offers a good first stab at an interesting problem: to look back in time at a web page. Their system performs relatively well for the scale that they implemented, but I don't know how useful it could be in reality. Their suite of tools are very sophisticated and truly help with interpreting the temporal data.

Aaron Hong - 10/31/2010 21:23:53

In "Edit Wear and Read Wear" Hill et al. talk about this system that visualizes document use, whether it's editing or reading. This is supposed to mimic real life physical wear of an object as it is used by humans. This goes along the theory of "reflective conversation" with the work materials, or also can be said as using a shared artifact over group process control.

I think that makes sense, especially the idea of mimicking wear to bring back some of the epistemic knowledge that we get from physical objects. To some degree we see limited implementations of this system, however not to the scope described in the paper. There are issues discussed, and from my limited knowledge, people haven't really looked into solving them.

In the second paper "Zoetrope: Interacting with the Ephemeral Web" by Adar et al. provide a interface for interacting with the ever changing web. They provide these timelines and lens that let you explore the space of ever updating websites. They also provide useful ways of visualizing the data and comparing across websites.

This is a particularly cool idea, and I would like to see some incarnation of this in a more mainstream implementation. The web changes a lot and I would like to be able to see the history of change. Like a movie or animation, the ability to scrub through the timeline provides invaluable access to information. In a similar way, the ability to scrub through the web's timeline would be useful also, in the case of making connections and realizations (like the gas price and news of war spendings example).