29 March 2005

Gillmor: Prioritizing aggregator displays based on user preferences

I was trying to wrap my brain around what Dan Gillmor was saying. After realizing I'd read everything and wasn't quite sure what he was saying, I decided on another approach.

I thought I'd contrast what he was saying with all the other ideas I've come up with, and try to make sense of things. Rather than sit on this, I thought I'd share what I was thinking while I was contrasting Gillmore's proposed standard ["Gillmore'] with the things already in my blog ["XML Concepts"]

XML Feed Splice takes a feed, and assigns your file plans to that feed. The idea is that you have your priorities and templates, and the trick is to assign weighting to the incoming feed. Gillmor's approach, on the other hand is not isolated to a single feed, but intended to apply to the entire aggregator.

One concept that is similar is the XML Ping Cluster which allows a reader to quickly ping those who are also commenting on the same blog. However, Gillmor is talking not just about a simple procedure, but a standard to prioritize these comments in the reader's aggregator. A ping cluster is about sending trackbacks quickly [a reader's response]; while the proposed standard is about reading quickly [not taking into consideration the response].

Also, the standard is similar to XML Novelty. However the distinctions are more than subtle. Gillmore's approach is to focus on the rankings as a means to identify value; while the Novelty approach does the opposite: It actively finds content that is least like what we might be expected to currently value.

XML Novelty is one that is linked more with searches for new content, not in screening existing content along an open channel. XML Novelty hopes to find new publication sources, while Gillmore hopes to do the opposite: Better filter the existing content from the open pipelines. One approach hopes to find new pipelines, while Gillmore is hoping to get more value through a thinner pipe, by getting rid of the sludge on the sides of the walls.

The standard also appears to be similar to XML Discussion Novelty. Gillmore's approach focuses on using this novelty to identify value, but goes one step further. Rather than simply flagging the novel items, Gillmore asks us to give a weighting to that novelty--it could be high, or it could be low.

Divergence is at the heart of the standard. Gillmore uses divergence as a basis to cut off discussion, or assign lower ratings. This doesn't mean that divergence is bad; rather, it simply accepts that divergence from one's defined parameters are less important or interesting than those that are more closely aligned. Then again, Gillmore does suggest that novel content can get a high ranking, if that is what the user prefers.

XML Discussion Support embraces the idea of divergence, but charts a course, rather than simply assigning lower values. The discussion support tool lets us focus on our goals, while Gillmore focuses on simply getting through the day's reading. How many times have we gotten lost in the meteor field on the way to the sun? [Uh...no comment.]

What I do like about what Gillmore's standard is his mechanism to evaluate the content we are reading. XML Discussion Signal does this. However, unlike the analysis of the feeds on a single platform, the discussion signal takes it to many levels and analyzes the discussions across multiple platforms.

One approach is self-focused while the other is multi-platform-conversation focused. Again, Gillmore hopes to simply better manage incoming information; the conversation cluster looks externally and hopes to manage many external conversations.

Duplicates: You may not read them, but they do provide information about the blogosphere's structure

One way of looking at the standard is thinking about the contrast between streamlining and aggregating. When we streamline, we taking the existing content and tailor it to our preferences. At the other end of the spectrum is the idea of taking everything and finding a new overarching structure to this content.

For example, we could look at how the duplicates are reviewed. The standard would focus on eliminating the duplicates. ON the other hand, duplicates could provide greater information about the information structure we are dealing with.

Another aspect of the idea is something about finding redundancies. This can be contrasted with XML Duplicate Analysis.

Gillmor's approach is to reduce the duplication; while XML Duplicate Analysis actually thrives on exploring the duplication. Gillmor wants to cut down on duplication to streamline reading; while the "XML Duplicate Analysis" does the opposite: Attracts and organizes the content related to a duplicate-source, into one single document.

Specifically, Gillmor's redundancy-reduction is a means to streamline a single feed; while the XML duplicate analysis focuses on a single document, and aggregates all the comments related to that content. One is streamlining to serve the busy reader; XML Duplicate Analysis is a consolidating and organizing mechanism for the content-searcher. The two are subtle, but the distinction drives different approaches to defining "what is a good XML platform."

Perhaps one of the initial tools under this standard might be something called a conversation cluster. Gillmor's explanation of the standard focuses on simply the receiver, while the conversation cluster takes everything and organizes it.

Another contrast is the design-structure. XML Link Maps creates a new external structure to many ideas; while Gillmor's approach turns the approach inside out by applying content to a pre-defined structure, the prioritizing and screening the content based on this structure. XML Link Maps take all the content and define a new structure; while Gillmor's approach takes the existing structure as defined by the user and applies it to the new content.

Again, Gillmor's approach screens out data. So does connecting content to one's objectives. However, Gillmor's approach is less connected with a single user-objective than with simply streamlining the existing data.

What Gillmor is proposing is going one step further than XML Feed Trace. Rather than simply tracing URLs to create a visual map for protecting information, Gillmor proposes using these common URLs to streamline information. XML Feed Trace is externally-focused, while Gillmor's approach is internally focused to one's own reading habits.

XML Format Assignment is probably the closest to what Gillmor is talking about. Essentially, both mechanisms hope to achieve the same objective -- tailor other's content to one's own objectives.

However, the two technical approaches, focus, and intent are entirely different. Gillmor's objective is again to simply expedite reading of a single aggregator; while the XML Format Assignment focuses on publishing and tracking on external platforms. Gillmor's approach is applying to the internal data and feeds we receive, while the XML Format Assignment is a way of taking user-defined preferences and quickly applying them to an external platform. One is for quick production, another is for quick consumption.

Gillmore also asks that we assign weights to content based on repuation. This is something covered in the XML Discussion Reputation. However, this tool and the others also listed in the same blog spot do not simply filter out information. These tools actively manage everything.

Whether users can read all the reports is another issue. Gillmore's approach more effectively tailors information, but does little to address the real issue that new information [stuff that doesn't appear important] could actually be the tip of something new. The trick is to know in advance how to recognize it, much less discover it.

There is a story about a drunken sailor on a random walk. To make a long story short, the story is a good one to teach statistics, and has little to do with sailing, drinking, or modern traveling.

Rather, it has more to do with how people find information when they are disorganized. Yet, there can be structured-disorganization, which can lead to many breakthroughs.

It's when we take the untrodden path, and can still figure out how to get back home in one piece that others are more likely to follow. Those dragons on the edge of Columbus' map were there to fill up space, not as a real message of what risks existed.

Back to the Gillmor discussion

After contrasting it with various XML Concepts, I have a better handle on what the standard is not. I'm struck with how similar the idea is to what Findory does: Takes a user's preferences and applies them to content.

I'd be interested in hearing more from Greg Linden in how "what Dan Gillmor is talking about" is unlike what already exists at Findory. I've got my own ideas, but I'd rather Greg weigh in.

It would be a shame if the XML-developer-wizards create something that already exists; at the same time, it would be a greater shame if a good idea got ignored simply because it sounded like something that should already exist. I look forward to reading more.

I was trying to wrap my brain around what Dan Gillmor was saying. After realizing I'd read everything and wasn't quite sure what he was saying, I decided on another approach.

I thought I'd contrast what he was saying with all the other ideas I've come up with, and try to make sense of things. Rather than sit on this, I thought I'd share what I was thinking while I was contrasting Gillmore's proposed standard ["Gillmore'] with the things already in my blog ["XML Concepts"]

XML Feed Splice takes a feed, and assigns your file plans to that feed. The idea is that you have your priorities and templates, and the trick is to assign weighting to the incoming feed. Gillmor's approach, on the other hand is not isolated to a single feed, but intended to apply to the entire aggregator.

One concept that is similar is the XML Ping Cluster which allows a reader to quickly ping those who are also commenting on the same blog. However, Gillmor is talking not just about a simple procedure, but a standard to prioritize these comments in the reader's aggregator. A ping cluster is about sending trackbacks quickly [a reader's response]; while the proposed standard is about reading quickly [not taking into consideration the response].

Also, the standard is similar to XML Novelty. However the distinctions are more than subtle. Gillmore's approach is to focus on the rankings as a means to identify value; while the Novelty approach does the opposite: It actively finds content that is least like what we might be expected to currently value.

XML Novelty is one that is linked more with searches for new content, not in screening existing content along an open channel. XML Novelty hopes to find new publication sources, while Gillmore hopes to do the opposite: Better filter the existing content from the open pipelines. One approach hopes to find new pipelines, while Gillmore is hoping to get more value through a thinner pipe, by getting rid of the sludge on the sides of the walls.

The standard also appears to be similar to XML Discussion Novelty. Gillmore's approach focuses on using this novelty to identify value, but goes one step further. Rather than simply flagging the novel items, Gillmore asks us to give a weighting to that novelty--it could be high, or it could be low.

Divergence is at the heart of the standard. Gillmore uses divergence as a basis to cut off discussion, or assign lower ratings. This doesn't mean that divergence is bad; rather, it simply accepts that divergence from one's defined parameters are less important or interesting than those that are more closely aligned. Then again, Gillmore does suggest that novel content can get a high ranking, if that is what the user prefers.

XML Discussion Support embraces the idea of divergence, but charts a course, rather than simply assigning lower values. The discussion support tool lets us focus on our goals, while Gillmore focuses on simply getting through the day's reading. How many times have we gotten lost in the meteor field on the way to the sun? [Uh...no comment.]

What I do like about what Gillmore's standard is his mechanism to evaluate the content we are reading. XML Discussion Signal does this. However, unlike the analysis of the feeds on a single platform, the discussion signal takes it to many levels and analyzes the discussions across multiple platforms.

One approach is self-focused while the other is multi-platform-conversation focused. Again, Gillmore hopes to simply better manage incoming information; the conversation cluster looks externally and hopes to manage many external conversations.

Duplicates: You may not read them, but they do provide information about the blogosphere's structure

One way of looking at the standard is thinking about the contrast between streamlining and aggregating. When we streamline, we taking the existing content and tailor it to our preferences. At the other end of the spectrum is the idea of taking everything and finding a new overarching structure to this content.

For example, we could look at how the duplicates are reviewed. The standard would focus on eliminating the duplicates. ON the other hand, duplicates could provide greater information about the information structure we are dealing with.

Another aspect of the idea is something about finding redundancies. This can be contrasted with XML Duplicate Analysis.

Gillmor's approach is to reduce the duplication; while XML Duplicate Analysis actually thrives on exploring the duplication. Gillmor wants to cut down on duplication to streamline reading; while the "XML Duplicate Analysis" does the opposite: Attracts and organizes the content related to a duplicate-source, into one single document.

Specifically, Gillmor's redundancy-reduction is a means to streamline a single feed; while the XML duplicate analysis focuses on a single document, and aggregates all the comments related to that content. One is streamlining to serve the busy reader; XML Duplicate Analysis is a consolidating and organizing mechanism for the content-searcher. The two are subtle, but the distinction drives different approaches to defining "what is a good XML platform."

Perhaps one of the initial tools under this standard might be something called a conversation cluster. Gillmor's explanation of the standard focuses on simply the receiver, while the conversation cluster takes everything and organizes it.

Another contrast is the design-structure. XML Link Maps creates a new external structure to many ideas; while Gillmor's approach turns the approach inside out by applying content to a pre-defined structure, the prioritizing and screening the content based on this structure. XML Link Maps take all the content and define a new structure; while Gillmor's approach takes the existing structure as defined by the user and applies it to the new content.

Again, Gillmor's approach screens out data. So does connecting content to one's objectives. However, Gillmor's approach is less connected with a single user-objective than with simply streamlining the existing data.

What Gillmor is proposing is going one step further than XML Feed Trace. Rather than simply tracing URLs to create a visual map for protecting information, Gillmor proposes using these common URLs to streamline information. XML Feed Trace is externally-focused, while Gillmor's approach is internally focused to one's own reading habits.

XML Format Assignment is probably the closest to what Gillmor is talking about. Essentially, both mechanisms hope to achieve the same objective -- tailor other's content to one's own objectives.

However, the two technical approaches, focus, and intent are entirely different. Gillmor's objective is again to simply expedite reading of a single aggregator; while the XML Format Assignment focuses on publishing and tracking on external platforms. Gillmor's approach is applying to the internal data and feeds we receive, while the XML Format Assignment is a way of taking user-defined preferences and quickly applying them to an external platform. One is for quick production, another is for quick consumption.

Gillmore also asks that we assign weights to content based on repuation. This is something covered in the XML Discussion Reputation. However, this tool and the others also listed in the same blog spot do not simply filter out information. These tools actively manage everything.

Whether users can read all the reports is another issue. Gillmore's approach more effectively tailors information, but does little to address the real issue that new information [stuff that doesn't appear important] could actually be the tip of something new. The trick is to know in advance how to recognize it, much less discover it.

There is a story about a drunken sailor on a random walk. To make a long story short, the story is a good one to teach statistics, and has little to do with sailing, drinking, or modern traveling.

Rather, it has more to do with how people find information when they are disorganized. Yet, there can be structured-disorganization, which can lead to many breakthroughs.

It's when we take the untrodden path, and can still figure out how to get back home in one piece that others are more likely to follow. Those dragons on the edge of Columbus' map were there to fill up space, not as a real message of what risks existed.

Back to the Gillmor discussion

After contrasting it with various XML Concepts, I have a better handle on what the standard is not. I'm struck with how similar the idea is to what Findory does: Takes a user's preferences and applies them to content.

I'd be interested in hearing more from Greg Linden in how "what Dan Gillmor is talking about" is unlike what already exists at Findory. I've got my own ideas, but I'd rather Greg weigh in.

It would be a shame if the XML-developer-wizards create something that already exists; at the same time, it would be a greater shame if a good idea got ignored simply because it sounded like something that should already exist. I look forward to reading more.

" />