During this morning’s IAC breakfast, discussing “Transparency, Collaboration and Web 2.0”, a panelist made a very interesting point. While the US Federal Government’s use of Internet social media services and cloud-based information-sharing applications is well underway, albeit at the very earliest of stages (mainly due to significant policy, privacy, security and simple “newness” issues), by far one of the major risks lies with accountability.
Accountability in collaboration and information-sharing environments is typically achieved to some degree by association of metadata with the information packages being exchanged, or with the “containers” of online events and the trusted identities of those participating in the dialogue. With social media applications and contexts like Twitter or Facebook, however, there are far too many ways that the “information packages” (i.e. unmanaged conversation bites) get exposed, syndicated and shared – disassociated from what I’ll call the “accountability metadata”. Accountability metadata might be described as one part records management (provenance, chain-of-custody, attribution, etc. of the actual material), one part situational awareness (i.e. the UCORE model; who, what, when, where regarding the actual event context being discussed), and one part “trust in context”, or “reputation” (i.e. popularity index, authority index, security attributes, etc.), and one part semantic accuracy (i.e. the topics and language being used is consistent with the context of discussion within which it’s introduced, for example according to a NIEM namespace).
"S-CORE" for "Social Media Core Information"?
As an example, the tweet associated with this blog entry is in fact an “information package”, albeit made up of unstructured data (as far as I can manipulate). Within the Twitter universe, there is an association (or “assertion”) of accountability metadata with this Tweet, so long as you’re a member of the community and can view my profile data, link references and prevalent topical themes. My profile data is associated with an employer and other communities, which themselves provide additional accountability data. But what happens when the Blog Tweet is Twitterfed, gets Tarpiped into Identica, over to Friendfeed, into Facebook and finally its RSS feed repurposed as a discussion item on someone else’s blog widget?
The original metadata isn’t carried along, and therefore some degree of manual intervention may be required to respond to non-attributed, out-of-context or otherwise mis-purposed data. Google searches may return search results containing my tweet language in non-intended contexts, thereby possibly enabling alternative or even incorrect interpretation. A homeland security social media “tweet”, for example, from a first-responder regarding a health-related assessment may be determined by HHS as inaccurate and possibly dangerous as most obviously interpreted by the public. Enter “Online Reputation Management”.
Online reputation management is a significant industry in itself (many local Washington DC Internet Marketing companies provide it), focused on making sure search engine results aren’t creating or promoting a false or unwarranted image – of a person, company, or product – because of overwhelming yet unverified online information posted to the contrary. Back to the HHS example – the erroneous tweet works its way into Google search results via multiple channels (and perhaps aggregate or federated search results), and subsequently becomes “the truth” because it’s on the first page of results. HHS or some other responsible, validating entity must now engage reputation management techniques to deliver more, better or different information into many of the same social media channels, in addition to a couple of its own highly-authoritative channels, to counter the ultimately false search engine results.
This may be part of the reason that Data.gov is so far a producer only of raw data, vs. “information” – since information carries with it expectations and actually delivers a degree of unstructured accountability that’s very hard to define, manage and monitor on the Internet. Structured data is far easier to manage, since it typically isn’t shared via social media (at least with its structure intact), and structured metadata is easily embedded. Citizens can certainly help resolve semantic inconsistencies, can establish some level of “social trust” by using the data, and can prove usefulness (and therefore legitimacy) by creating popular applications – but citizens aren’t really accountable to the rest of the Federal Government’s constituency, and its reputation. Ultimately, organizations will look back to the Government for trust and accountability with respect to information packages (vs. data) they can use in legitimate business ventures involving social media.
However, the Federal Government’s foray into producing information packages for consumption by public social media will likely be constrained for some to come, until industry can come up with a generally accepted standard and technology examples to permanently associate “accountability metadata” to unstructured information payloads released in the wild. This might then be followed by an oversight agency or program that could automate perhaps some of the Federal Reputation Management tasks that would then be necessary – enabling many more useful, unstructured conversations in public social media, moderated by trusted Government sources. Perhaps from the cloud.
Accountability in collaboration and information-sharing environments is typically achieved to some degree by association of metadata with the information packages being exchanged, or with the “containers” of online events and the trusted identities of those participating in the dialogue. With social media applications and contexts like Twitter or Facebook, however, there are far too many ways that the “information packages” (i.e. unmanaged conversation bites) get exposed, syndicated and shared – disassociated from what I’ll call the “accountability metadata”. Accountability metadata might be described as one part records management (provenance, chain-of-custody, attribution, etc. of the actual material), one part situational awareness (i.e. the UCORE model; who, what, when, where regarding the actual event context being discussed), and one part “trust in context”, or “reputation” (i.e. popularity index, authority index, security attributes, etc.), and one part semantic accuracy (i.e. the topics and language being used is consistent with the context of discussion within which it’s introduced, for example according to a NIEM namespace).
"S-CORE" for "Social Media Core Information"?
As an example, the tweet associated with this blog entry is in fact an “information package”, albeit made up of unstructured data (as far as I can manipulate). Within the Twitter universe, there is an association (or “assertion”) of accountability metadata with this Tweet, so long as you’re a member of the community and can view my profile data, link references and prevalent topical themes. My profile data is associated with an employer and other communities, which themselves provide additional accountability data. But what happens when the Blog Tweet is Twitterfed, gets Tarpiped into Identica, over to Friendfeed, into Facebook and finally its RSS feed repurposed as a discussion item on someone else’s blog widget?
The original metadata isn’t carried along, and therefore some degree of manual intervention may be required to respond to non-attributed, out-of-context or otherwise mis-purposed data. Google searches may return search results containing my tweet language in non-intended contexts, thereby possibly enabling alternative or even incorrect interpretation. A homeland security social media “tweet”, for example, from a first-responder regarding a health-related assessment may be determined by HHS as inaccurate and possibly dangerous as most obviously interpreted by the public. Enter “Online Reputation Management”.
Online reputation management is a significant industry in itself (many local Washington DC Internet Marketing companies provide it), focused on making sure search engine results aren’t creating or promoting a false or unwarranted image – of a person, company, or product – because of overwhelming yet unverified online information posted to the contrary. Back to the HHS example – the erroneous tweet works its way into Google search results via multiple channels (and perhaps aggregate or federated search results), and subsequently becomes “the truth” because it’s on the first page of results. HHS or some other responsible, validating entity must now engage reputation management techniques to deliver more, better or different information into many of the same social media channels, in addition to a couple of its own highly-authoritative channels, to counter the ultimately false search engine results.
This may be part of the reason that Data.gov is so far a producer only of raw data, vs. “information” – since information carries with it expectations and actually delivers a degree of unstructured accountability that’s very hard to define, manage and monitor on the Internet. Structured data is far easier to manage, since it typically isn’t shared via social media (at least with its structure intact), and structured metadata is easily embedded. Citizens can certainly help resolve semantic inconsistencies, can establish some level of “social trust” by using the data, and can prove usefulness (and therefore legitimacy) by creating popular applications – but citizens aren’t really accountable to the rest of the Federal Government’s constituency, and its reputation. Ultimately, organizations will look back to the Government for trust and accountability with respect to information packages (vs. data) they can use in legitimate business ventures involving social media.
However, the Federal Government’s foray into producing information packages for consumption by public social media will likely be constrained for some to come, until industry can come up with a generally accepted standard and technology examples to permanently associate “accountability metadata” to unstructured information payloads released in the wild. This might then be followed by an oversight agency or program that could automate perhaps some of the Federal Reputation Management tasks that would then be necessary – enabling many more useful, unstructured conversations in public social media, moderated by trusted Government sources. Perhaps from the cloud.
Comments