Information granularity

How big is information anyway?

To answer this, some researchers (Jones, Keeping found things found, 2008) have defined the information item as:

information encapsulated in in a persistent form that can be created, modified, stored, retrieved, given a name. tags, and other properties, moved, copied, distributed, deleted or otherwise manipulated. And information item has an associated information form that is determined by the tools and applications that support these operations.

Information item can be a paper document (tools are paper clips, file cabinets …), files like documents, photos, videos and other file types (applications like file, photo, video managers …), emails (email clients), web sites and shortcuts or references to any of these.

What about our conversations? A phone conversation is not an information item. Only if we tape it, the recorded sound stream becomes an information item. What about other conversations (asynchronous and synchronous)?

In IMs (instant messaging), chats are recorded in a chat history. A chat history can be viewed, copied, deleted … But some IM applications store these conversation by contacts and a history of one contact includes many conversations (to be precise – all of them)! What is now an information item: all conversations together (as they are packed in a file that can be manipulated) or every conversation separately. Users certainly see these conversations as separated entities. Computer does not. It separates them by time tags.

What about email messages. We can say that an email seen by an email client as an entity is information item (corresponding to the above definition). But sometimes conversations span over several email messages (located in separate folders like in Inbox and Sent folder as well). Now this conversation forms an information package as a whole, but it is split over several information items (according to the definition). (OK, some email clients can show email threads!)

Let’s consider an old media: a tape or a music cassette. By definition it is an information item. But users certainly didn’t necessarily always see it like that. They might have bought it for one or two songs they liked and never listened to the rest of it!? What is information item to them? More to the present: a user bookmarks a web site spanning over several computer screens because of a little tiny part of it (let say a paragraph). What is information item for this user? The whole site or just a paragraph that was the reason for the site to be bookmarked in the first place?

Granularity of information is different for users and computers. People cut articles out of newspapers they are interested in, and don’t save the whole pile of newspapers. It is unreasonable saving a pile for just a few scraps. People don’t have the possibility to bookmark only a part of a website (there are few tools like Snip!t that can help). Most people don’t know how  to separate chat histories (except for copying and pasting) or have the knowledge to split an audio file. But do we need such tools and such knowledge? Is packaging information in the present persistent forms right for our needs?

EDIT 07/06/2010:

Gemmell (MyLifeBits: Fulfilling the Memex vision, 2002) argued that links between infromation are important as they help users find context and comentary: "When a quote is given or a video clip out of a longer video used, the pointer back to the original material answers the question: what is the context this was taken from". Similarly Karger (Data unification in PIM) states that copying and pasting information loses its original context of the original information item (and items in the same category). Haystack (Huynh, Haystack: a platform for creating, organizing and visualizing semistructured information, 2003) made it possible to group, annotate and link at smaller and more meaningful units than a file. ProjectPlanner also allows to use only parts of information from web sites emails and files that could be dragged and dropped to a word processor environment for planning projects (Jones et. al., The personal project planner: planning to organize personal information, 2008).