The recent riots in Kampala have revealed a couple of things about the flow of news and how people behave in crisis situations. We still don’t quite know what we’re dealing with in Kampala. It’s either the beginning or the end of a wider scale confrontation. For those of us on the ground, we’re starved for information. The mobile phone and it’s users ere essentially the only reports we can rely on for timely info. Citizen journalists like Solomon risked arrest and their personal safety by literally reporting from the ‘red zone’.
Despite the fact that people are more connected than ever through mobile devices and web services like Twitter, there are still some gaping holes in how information is aggregated and disseminated in times of crisis. Over the past three days I contemplated how it might be possible to improve the flow of data during crisis.
Real Time Reporting from Verified Sources
Two days ago the data flow was sparse, mainstream coverage was disjointed. When we did get updates it was essentially “the rioting is still going on, here’s why it started and what’s being done to stop it”. But to be fair, there’s no way media outlets can provide the type of real-time updates we all need on the ground without having people on the ground reporting. It would make sense for the news outlets to aggregate the reports of local journalists like Ugandan Insomniac or The Independent’s @UgandaTalks, who did attempt to offer realtime coverage in their official roles as journalists.
SMS is like the new short-wave radio but it can be hindered (due to high traffic to networks or provider shut-down). Still, it’s how we spread the news of our own situation as it unfolded. However, if the news industry can’t rely on real-time citizen reports, then they need to rely on real-time reports from embedded reporters. In this way reporters like @UgandaTalks acted as sort of an authoritative filter. We know he’s a journalist working for The Independent, we know he’s on the ground. He knows some of the citizen journalists and forwarded (or re-tweeted) a lot of what they were reporting while also offering his own perspective. While the major news outlets of Uganda also turned to Twitter, most used the service differently, either to push non-realtime articles or to offer sporadic updates. The bigger problem here is that when there were indeed unverified reports of a ‘media blackout’ they couldn’t offer any context at all. Why? Because they are ‘big media’ and they were most likely the ones ‘blacked out’. Their silence only verified those initial reports for some and made everyone else suspicious.
This is where an application like Ushahidi shines, in the collection of reports and contextualization of them with qualitative data (verified versus unverified, identifying bloggers versus news outlets etc.) The pictures from ugandawitness.com below show how Ushahidi works, aggregating reports of incident from Twitter, the news and the greater blogosphere and mapping them.
To the left reports of violence, to the right aggregated blogs, articles and other media.
Additional Context (Other Reports, History, Related Events)
When situations like this occur, for most people in the world, they couldn’t be any later to turn their attention to the situation (by the time violence erupts, it usually means the cause has been brewing for some time). What most people need is a way to aggregate background information on a subject in a more ‘digestible’ form. Simply aggregating messages from the people in crisis does nothing to contextualize this info for them. They also need the ‘why?’
The BBC has done an excellent job of revamping their article pages to do this as the image below from their coverage of the 2008 Mumbai Attacks shows.
To the left we have pictures, the right lists recent related articles as well as background analysis and eye witness accounts from the ground. Fairly simple, but very useful for the average viewer who’ll be new to the whole situation. Where I think this is headed is to the ‘wiki-fication’ of news, where some info is coverage from professional reporters, while the rest is ‘user generated’ but curated (edited) by professionals.
Chronological Attribution
Another contextual clue that’s missing from distributed micro-messaging tools like Twitter is that of time. I suppose it’s not missing, it just often gets over looked.
On day three of the Kampala riots we saw evidence of this. People on the ground wrote about the situation as it unfolded through out the day until things began to finally settle down. Then 8 hours later we began to see people in the US retweeting things that happened hours ago (in the morning for us) as ‘news’. To them it was indeed news as they were just waking up to it. This in turn created confusion for us back here. (“Have the riots started again or is that just an old re-tweet?”). This is not a problem of user behavior, rather it’s a problem of the UI itself. Retweeting messages from friends and followers is something users of Twitter have always done. However, what’s missing is some way to group these updates together, so as to build a more accurate timeline of activity. On services like Friendfeed when identical messages are posted, the service automatically groups them in the users stream. It’d be interested to see this applied to Twitter’s stream of updates.
At a glance now we can see that we have several messages that are so similar that they are probably identical, or at the very least related. In the case of Twitter, any and all ‘retweets’ could simply be grouped as they’re essentially all reverberations of that one initial event that we can think of as the stimulus. (ex. My twitter handle is @jongos, I send the message “fire!”, it becomes the stimulus and thus any message that says “r/t @jongos fire!” or “fire (via @jongos”) is just an echo.)
It would be more useful to have all related tweets collapsable. Additionally, it would be useful if there were a way to view these happenings with some sort of chronological barometer. If shifted along a horizontal axis, like a timeline, we’d end up with something that looks a bit like the social media application Plurk.
With Plurk the updates appear along a horizontal grid that has temporal indicators. An update at 1:14am will appear farther left than one at 1:19am. As time progresses the data flows from right to left (so as to be read from left to right).
Curated Media as Context
When stories of the Kampala riots first broke, a number of people began looking for photos. However the photos they found were old, although they were indeed from Kampala riots, they were from riots that took place in 2005. So adding media like photos and videos is incredibly important. Anyone should be able to submit data, while a few select editors would curate and publish it. Additionally, each photo, video or audio recording should have some additional contextual information (location data, description, transcript, timestamp and related media). A good example of these in use was http://voteindia.in.
Misinformation, Inaccuracies and Speculation
As much as there needs to be a way to add contextual info about the history of a situation, there needs to be a way to aggregate rumors, inaccuracies and speculative comments. For an example, this one, posted yesterday during the riots:
@WomenCan seems to have been misinformed about the situation and the reasons behind what was occurring. I don’t think there’s a person in Uganda who would have called last weeks riots ‘faith based’. Similarly, trying to sum it all up as a ‘land dispute’ (like a few western media outlets did) isn’t entirely accurate either. Misinformation like this needs to be aggregated so that it can be clarified or authoritatively disproven and offered back to the public as a resource to learn from. Even more importantly, linking the inaccurate sources so that the erroneous publisher might become aware of their mistakes (via trackbacks or comments left on their blogs) is also necessary.
Applications like Swiftriver (pictured below) are attempting to do this.
Of course, as with any discussion, there will be ongoing discussion regarding facts and people who will disagree. Much-like wikipedia’s ‘discussion’ area, there needs to be a running conversation allowing the discussions to happen as they can offer valuable insight. Whether it’s through some sort of voting mechanism or via trusted sources, there needs to be a way of filtering conflicting reports while not completely suppressing them, after all dissent can offer additional insight.
The Snowball Effect
During the recent conflicts in Georgia and Iran, the sheer number of reports coming from the region catapulted the stories into the mainstream news. This is called the ‘snowball effect’ as eventually so many people are reporting on something that it begs additional attention (like a snowball picking up more snow as it rolls downhill). In the case of Kampala, because there were so few people reporting via blogs or Twitter about the riots, and subsequently there was less attention. What seemed like a lifeline of information to some, went completely oblivious to others. This was also the case during the 2009 Iranian Election Crisis. Still, the very fact that it’s possible for citizens to push the media’s attention when it needs to be is valuable.
In this case, the fact that the main stream news was late to pick up on the unfolding situation and that it went unnoticed by Twitter’s trending utility just highlights how important it is for locals to rely on each other for reports.
Realtime Search
Search became incredibly important as well. Most people following the reports on Twitter used various real-time search engines filtered by the word ‘kampala’. This allowed us to simply refresh one window to get all the latest updates. Rather than trying to look through all our various friends tweets. Although Facebook is the primary way Ugandans socialize on the web, the lack of an open search API really hindered the services usefulness. We all know Facebook more open search is indeed coming but it needs to allow for search beyond just a user’s network of friends to be useful.
Are there other ways in which crisis reporting can be improved? Tell us in the comments below.
Asynchronous Info, Disjointed Data and Crisis Reporting
The recent riots in Kampala have revealed a couple of things about the flow of news and how people behave in crisis situations. We still don’t quite know what we’re dealing with in Kampala. It’s either the beginning or the end of a wider scale confrontation. For those of us on the ground, we’re starved for information. The mobile phone and it’s users ere essentially the only reports we can rely on for timely info. Citizen journalists like Solomon risked arrest and their personal safety by literally reporting from the ‘red zone’.
Despite the fact that people are more connected than ever through mobile devices and web services like Twitter, there are still some gaping holes in how information is aggregated and disseminated in times of crisis. Over the past three days I contemplated how it might be possible to improve the flow of data during crisis.
Real Time Reporting from Verified Sources
Two days ago the data flow was sparse, mainstream coverage was disjointed. When we did get updates it was essentially “the rioting is still going on, here’s why it started and what’s being done to stop it”. But to be fair, there’s no way media outlets can provide the type of real-time updates we all need on the ground without having people on the ground reporting. It would make sense for the news outlets to aggregate the reports of local journalists like Ugandan Insomniac or The Independent’s @UgandaTalks, who did attempt to offer realtime coverage in their official roles as journalists.
SMS is like the new short-wave radio but it can be hindered (due to high traffic to networks or provider shut-down). Still, it’s how we spread the news of our own situation as it unfolded. However, if the news industry can’t rely on real-time citizen reports, then they need to rely on real-time reports from embedded reporters. In this way reporters like @UgandaTalks acted as sort of an authoritative filter. We know he’s a journalist working for The Independent, we know he’s on the ground. He knows some of the citizen journalists and forwarded (or re-tweeted) a lot of what they were reporting while also offering his own perspective. While the major news outlets of Uganda also turned to Twitter, most used the service differently, either to push non-realtime articles or to offer sporadic updates. The bigger problem here is that when there were indeed unverified reports of a ‘media blackout’ they couldn’t offer any context at all. Why? Because they are ‘big media’ and they were most likely the ones ‘blacked out’. Their silence only verified those initial reports for some and made everyone else suspicious.
This is where an application like Ushahidi shines, in the collection of reports and contextualization of them with qualitative data (verified versus unverified, identifying bloggers versus news outlets etc.) The pictures from ugandawitness.com below show how Ushahidi works, aggregating reports of incident from Twitter, the news and the greater blogosphere and mapping them.
To the left reports of violence, to the right aggregated blogs, articles and other media.
Additional Context (Other Reports, History, Related Events)
When situations like this occur, for most people in the world, they couldn’t be any later to turn their attention to the situation (by the time violence erupts, it usually means the cause has been brewing for some time). What most people need is a way to aggregate background information on a subject in a more ‘digestible’ form. Simply aggregating messages from the people in crisis does nothing to contextualize this info for them. They also need the ‘why?’
The BBC has done an excellent job of revamping their article pages to do this as the image below from their coverage of the 2008 Mumbai Attacks shows.
To the left we have pictures, the right lists recent related articles as well as background analysis and eye witness accounts from the ground. Fairly simple, but very useful for the average viewer who’ll be new to the whole situation. Where I think this is headed is to the ‘wiki-fication’ of news, where some info is coverage from professional reporters, while the rest is ‘user generated’ but curated (edited) by professionals.
Chronological Attribution
Another contextual clue that’s missing from distributed micro-messaging tools like Twitter is that of time. I suppose it’s not missing, it just often gets over looked.
On day three of the Kampala riots we saw evidence of this. People on the ground wrote about the situation as it unfolded through out the day until things began to finally settle down. Then 8 hours later we began to see people in the US retweeting things that happened hours ago (in the morning for us) as ‘news’. To them it was indeed news as they were just waking up to it. This in turn created confusion for us back here. (“Have the riots started again or is that just an old re-tweet?”). This is not a problem of user behavior, rather it’s a problem of the UI itself. Retweeting messages from friends and followers is something users of Twitter have always done. However, what’s missing is some way to group these updates together, so as to build a more accurate timeline of activity. On services like Friendfeed when identical messages are posted, the service automatically groups them in the users stream. It’d be interested to see this applied to Twitter’s stream of updates.
At a glance now we can see that we have several messages that are so similar that they are probably identical, or at the very least related. In the case of Twitter, any and all ‘retweets’ could simply be grouped as they’re essentially all reverberations of that one initial event that we can think of as the stimulus. (ex. My twitter handle is @jongos, I send the message “fire!”, it becomes the stimulus and thus any message that says “r/t @jongos fire!” or “fire (via @jongos”) is just an echo.)
It would be more useful to have all related tweets collapsable. Additionally, it would be useful if there were a way to view these happenings with some sort of chronological barometer. If shifted along a horizontal axis, like a timeline, we’d end up with something that looks a bit like the social media application Plurk.
With Plurk the updates appear along a horizontal grid that has temporal indicators. An update at 1:14am will appear farther left than one at 1:19am. As time progresses the data flows from right to left (so as to be read from left to right).
Curated Media as Context
When stories of the Kampala riots first broke, a number of people began looking for photos. However the photos they found were old, although they were indeed from Kampala riots, they were from riots that took place in 2005. So adding media like photos and videos is incredibly important. Anyone should be able to submit data, while a few select editors would curate and publish it. Additionally, each photo, video or audio recording should have some additional contextual information (location data, description, transcript, timestamp and related media). A good example of these in use was http://voteindia.in.
Misinformation, Inaccuracies and Speculation
As much as there needs to be a way to add contextual info about the history of a situation, there needs to be a way to aggregate rumors, inaccuracies and speculative comments. For an example, this one, posted yesterday during the riots:
@WomenCan seems to have been misinformed about the situation and the reasons behind what was occurring. I don’t think there’s a person in Uganda who would have called last weeks riots ‘faith based’. Similarly, trying to sum it all up as a ‘land dispute’ (like a few western media outlets did) isn’t entirely accurate either. Misinformation like this needs to be aggregated so that it can be clarified or authoritatively disproven and offered back to the public as a resource to learn from. Even more importantly, linking the inaccurate sources so that the erroneous publisher might become aware of their mistakes (via trackbacks or comments left on their blogs) is also necessary.
Applications like Swiftriver (pictured below) are attempting to do this.
Of course, as with any discussion, there will be ongoing discussion regarding facts and people who will disagree. Much-like wikipedia’s ‘discussion’ area, there needs to be a running conversation allowing the discussions to happen as they can offer valuable insight. Whether it’s through some sort of voting mechanism or via trusted sources, there needs to be a way of filtering conflicting reports while not completely suppressing them, after all dissent can offer additional insight.
The Snowball Effect
During the recent conflicts in Georgia and Iran, the sheer number of reports coming from the region catapulted the stories into the mainstream news. This is called the ‘snowball effect’ as eventually so many people are reporting on something that it begs additional attention (like a snowball picking up more snow as it rolls downhill). In the case of Kampala, because there were so few people reporting via blogs or Twitter about the riots, and subsequently there was less attention. What seemed like a lifeline of information to some, went completely oblivious to others. This was also the case during the 2009 Iranian Election Crisis. Still, the very fact that it’s possible for citizens to push the media’s attention when it needs to be is valuable.
In this case, the fact that the main stream news was late to pick up on the unfolding situation and that it went unnoticed by Twitter’s trending utility just highlights how important it is for locals to rely on each other for reports.
Realtime Search
Search became incredibly important as well. Most people following the reports on Twitter used various real-time search engines filtered by the word ‘kampala’. This allowed us to simply refresh one window to get all the latest updates. Rather than trying to look through all our various friends tweets. Although Facebook is the primary way Ugandans socialize on the web, the lack of an open search API really hindered the services usefulness. We all know Facebook more open search is indeed coming but it needs to allow for search beyond just a user’s network of friends to be useful.
Are there other ways in which crisis reporting can be improved? Tell us in the comments below.