Can We Rely On Social Media In An Emergency?

15 views
Skip to first unread message

Jon Gos

unread,
Jan 13, 2012, 6:26:04 PM1/13/12
to swift...@googlegroups.com
...While there is no denying that Facebook, Twitter and other social networks help spread critical information when emergencies strike, they can still be problematic and ineffective when compared to other forms of communication. Even the Bangkok terror alert was met with initial skepticism, and while the embassy has close to 40,000 Twitter followers, many of those were not in the area of the threat...


--
Jonathan D. Gosier // TED Host
Co-Founder, metaLayer Inc.

Charl van Niekerk

unread,
Jan 14, 2012, 12:22:09 PM1/14/12
to swift...@googlegroups.com
Extremely interesting stuff, Jon. It would have been really cool if
Twitter had some kind of "emergency broadcast" system that would
spread critical (curated) information to all its users within a
certain geographical area. Tools like SwiftRiver could also be used to
fill this void; some interesting possibilities to explore here.

Jon Gosier

unread,
Jan 14, 2012, 12:30:32 PM1/14/12
to swift...@googlegroups.com
Great idea. Ushahidi and Swift are basically that (if you look at it from a certain perspective). I wonder if there's an opportunity to offer curated, relevant, information during crisis events back to Twitter, which they could then target using their 'promoted tweet' advertising platform.

--
Jonathan D. Gosier // TED Senior Fellow

skype | j.gosier





--
You received this message because you are subscribed to the Google Groups "SwiftRiver" group.
To post to this group, send email to swift...@googlegroups.com.
To unsubscribe from this group, send email to swiftriver+...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/swiftriver?hl=en.



Charl van Niekerk

unread,
Jan 14, 2012, 12:36:33 PM1/14/12
to swift...@googlegroups.com
2012/1/14 Jon Gosier <j...@swiftly.org>:

> Great idea. Ushahidi and Swift are basically that (if you look at it from a
> certain perspective). I wonder if there's an opportunity to offer curated,
> relevant, information during crisis events back to Twitter, which they could
> then target using their 'promoted tweet' advertising platform.

Brilliant thinking. I like where this is going.

Alternatively, we could also look at building mobile applications that
serve purely as emergency notifiers. Never used an iPhone but the
Android has a notification area that slides down from the top - the
app itself can be minimalist, just dropping notifications in there. Of
course, this needs to be location aware, which shouldn't be a problem
on any of today's modern smart phones.

We could even potentially build it on top of this:

http://notifo.com/

Ali-Reza Anghaie

unread,
Jan 14, 2012, 12:36:32 PM1/14/12
to swift...@googlegroups.com
I think this is a great idea on paper but I'd be afraid of formerly
politicizing ~any~ network for this purpose. Let the ppl RT it to
their hearts content but I view any such formalized "promoted" effort
to only be supporting the idea that some Governments have that this is
a valid Government use. Or making Twitter (and friends) further
targets of hacktivism groups.

Just my $0.02 and exchange rates vary. -Ali

Charl van Niekerk

unread,
Jan 14, 2012, 12:38:44 PM1/14/12
to swift...@googlegroups.com
2012/1/14 Ali-Reza Anghaie <a...@packetknife.com>:

> I think this is a great idea on paper but I'd be afraid of formerly
> politicizing ~any~ network for this purpose. Let the ppl RT it to
> their hearts content but I view any such formalized "promoted" effort
> to only be supporting the idea that some Governments have that this is
> a valid Government use. Or making Twitter (and friends) further
> targets of hacktivism groups.
>
> Just my $0.02 and exchange rates vary. -Ali

Valid concern, but I think the idea is to keep this politically
neutral. Knowledge of a possible bomb threat is not politically
polarising in any way IMHO.

Ali-Reza Anghaie

unread,
Jan 14, 2012, 12:39:24 PM1/14/12
to swift...@googlegroups.com
Don't neglect SMS notification systems or various amateur or shortwave
systems. Especially given some of the target regions of hotbed
activity recently. -Ali

Ali-Reza Anghaie

unread,
Jan 14, 2012, 12:41:44 PM1/14/12
to swift...@googlegroups.com
I'm looking forward to seeing the new Beta roll-out. See what APIs or
RESTful stuff is available, because there are a good number of ways to
leverage the information that augment the platform's already beautiful
work.

It's not wrong to get a little giddy at the possibilities of good
wholesome work. ;-)

-Ali

Charl van Niekerk

unread,
Jan 14, 2012, 12:43:13 PM1/14/12
to swift...@googlegroups.com
2012/1/14 Ali-Reza Anghaie <a...@packetknife.com>:

> Don't neglect SMS notification systems or various amateur or shortwave
> systems. Especially given some of the target regions of hotbed
> activity recently. -Ali

SMS is always the optimal solution as it works on any device (except
it is limited in what it can carry - only short text messages, as the
name suggests). The problem of cost will always remain however.

I am a big fan of HAM radio, although I've never gotten involved with
it myself. South Africa has a very healthy HAM radio community:

http://www.sarl.org.za/

Packet radio is also an interesting subject:

http://en.wikipedia.org/wiki/Packet_radio

Ali-Reza Anghaie

unread,
Jan 14, 2012, 12:50:27 PM1/14/12
to swift...@googlegroups.com
Agreed. The Swiftriver BBS network on the other end of packet radio.
Man, that brings back memories on two fronts.

And as a last plug I think the P2P ad-hoc solutions like
https://fluidnexus.net/ should certainly be on the list.

See y'all on the other side of beta. -Ali

Charl van Niekerk

unread,
Jan 14, 2012, 3:11:37 PM1/14/12
to swift...@googlegroups.com
2012/1/14 Ali-Reza Anghaie <a...@packetknife.com>:

> And as a last plug I think the P2P ad-hoc solutions like
> https://fluidnexus.net/ should certainly be on the list.

Most definitely! When the government kills the network, this will be a
necessity.

Nick Arnett

unread,
Jan 15, 2012, 11:44:51 AM1/15/12
to swift...@googlegroups.com
This is not a simple question at all.  For one thing, there are various scales of emergency.  Social media *is* working well in some kinds of emergencies, for people to self-organization, for agencies to get information out, etc.  These are smaller incidents where the affected population is social-media savvy, generally speaking.  

As far as gathering intelligence via social media during an incident is concerned, it depends on who is doing the gathering and for what purpose.  I was able to support a team in Ishinomaki after the Japan earthquake/tsumani with a fair bit of actionable information I gathered mostly from Twitter - where the shelters were, where water and electricity were available, etc., just a few days after the event.  But it was a painfully manual process and was effective in no small part due to the fact that I knew the team and could communicate directly with them.  We had been in Haiti together, so I knew what kind of info would be most valuable and they knew that they could rely on me as a source.

If we're considering whether or not U.S. first responder agencies are ready to rely on social media for assessment, response, etc., we're a long way from there.  There's high skepticism about the reliability of information from social media and when incidents are large, it is very difficult to remove duplicates, identify locations, the actual nature of what's needed, etc.

In the event of terrorist attack, information from social media has been used to target follow-up attacks, which will make government reluctant to encourage people to share data that indicates where crowds are gathered, where police are responding or not, etc.  

All of that aside, this is the area I'm focused on when I'm not doing my regular job, which is director of product management for one of the biggest social media analytics companies. I'd like to try hooking our system into SwiftRiver, but the last time I tried to get SwiftRiver running, I couldn't, even on a cleanly installed server, and didn't get much help... is it better now?

Nick

Charl van Niekerk

unread,
Jan 15, 2012, 4:30:44 PM1/15/12
to swift...@googlegroups.com
2012/1/15 Nick Arnett <nick....@gmail.com>:

> All of that aside, this is the area I'm focused on when I'm not doing my
> regular job, which is director of product management for one of the biggest
> social media analytics companies. I'd like to try hooking our system into
> SwiftRiver, but the last time I tried to get SwiftRiver running, I couldn't,
> even on a cleanly installed server, and didn't get much help... is it better
> now?

Interesting questions, Nick. Regarding the issue of support - I like
to think this mailing list has provided a decent channel for support
for all of our products, but which version of which product in
particular are you referring to?

Nick Arnett

unread,
Jan 15, 2012, 5:23:39 PM1/15/12
to swift...@googlegroups.com
On Sun, Jan 15, 2012 at 1:30 PM, Charl van Niekerk <ch...@ushahidi.com> wrote:

Interesting questions, Nick. Regarding the issue of support - I like
to think this mailing list has provided a decent channel for support
for all of our products, but which version of which product in
particular are you referring to?

Let's see... I first tried to get an instance working at CMU the day after the Japan quake, which was mid-March, then tried again a couple of weeks later.  I think I was trying to get Sweeper working, but the install, on a clean install of Ubuntu, just wouldn't complete.

Hard to recall, exactly, after this long.

I'd love to try plugging our API into SwiftRiver, but don't have a lot of time to put into it.  We could stream a large amount of content with deep natural language tagging into it.  I'd especially love to see how the workflow components might be able to take advantage of our tags.  I've talked with our major data providers about making content available at no extra cost for disaster response and they're quite agreeable.  Actually, to be clear, there is only one provider whose pricing is volume-dependent, but they are quite supportive.

Perhaps it's time to try again... has anybody ever integrated a feed that contains tags of things like named entities, behaviors, emotions, locations, etc.?

Nick

Charl van Niekerk

unread,
Jan 16, 2012, 4:48:31 AM1/16/12
to swift...@googlegroups.com
2012/1/15 Nick Arnett <nick....@gmail.com>:

> Let's see... I first tried to get an instance working at CMU the day after
> the Japan quake, which was mid-March, then tried again a couple of weeks
> later.  I think I was trying to get Sweeper working, but the install, on a
> clean install of Ubuntu, just wouldn't complete.

Not sure what exactly went wrong, but I used to run many deployments
on Debian / Ubuntu without any trouble. Even wrote a script to
automate the process:

https://github.com/ushahidi/Sweeper_0.3.2/blob/master/deploy/debian.sh

Please note that, as per the README, Sweeper has been deprecated in
anticipation of the new SwiftRiver:

http://blog.ushahidi.com/index.php/2011/12/13/swiftriver-throws-a-lifeline-to-people-drowning-in-information/

> I'd love to try plugging our API into SwiftRiver, but don't have a lot of
> time to put into it.  We could stream a large amount of content with deep
> natural language tagging into it.  I'd especially love to see how the
> workflow components might be able to take advantage of our tags.  I've
> talked with our major data providers about making content available at no
> extra cost for disaster response and they're quite agreeable.  Actually, to
> be clear, there is only one provider whose pricing is volume-dependent, but
> they are quite supportive.
>
> Perhaps it's time to try again... has anybody ever integrated a feed that
> contains tags of things like named entities, behaviors, emotions, locations,
> etc.?

Nothing particularly springs to mind, but my best suggestion would be
to wait for the release of the new beta before proceeding, then
building directly on the new system.

Best Regards,
Charl

George Weyman

unread,
Jan 16, 2012, 8:40:46 AM1/16/12
to swift...@googlegroups.com
Broadcasting 'threats' can definitely be polititcal - the US colour coded threat level system for example, or the UK government's deployment of tanks at heathrow http://news.bbc.co.uk/1/hi/uk/2749659.stm.  The public never has access to the critical information (if it exists) that triggered the warning.

It might work if there were very narrow categories of use incorporated into government policy, ie. events that have happened and that threaten the health and well-being of local populations, for example a certain incidence of water borne disease in your area.  But even then it would surely need to be something you opt into.

Cheers for the discussion!
-George



--
You received this message because you are subscribed to the Google Groups "SwiftRiver" group.
To post to this group, send email to swift...@googlegroups.com.
To unsubscribe from this group, send email to swiftriver+...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/swiftriver?hl=en.




--
George Weyman  - Senior Program Officer, Meedan |
+44 7792 921298 +44 203 0155306 |
Meedan in Wired - http://bit.ly/Meedan-in-Wired |
Meedan Explained in 60 Seconds - http://bit.ly/Meedan-in-60Seconds  |
Meedan on Twitter - twitter.com/meedan | Meedan Blog – http://meedan.org/blog/ |


Charl van Niekerk

unread,
Jan 16, 2012, 9:11:58 AM1/16/12
to swift...@googlegroups.com
2012/1/16 George Weyman <gwe...@meedan.net>:

> Broadcasting 'threats' can definitely be polititcal - the US colour coded
> threat level system for example, or the UK government's deployment of tanks
> at heathrow http://news.bbc.co.uk/1/hi/uk/2749659.stm.  The public never has
> access to the critical information (if it exists) that triggered the
> warning.
>
> It might work if there were very narrow categories of use incorporated into
> government policy, ie. events that have happened and that threaten the
> health and well-being of local populations, for example a certain incidence
> of water borne disease in your area.  But even then it would surely need to
> be something you opt into.

Yeah but what I meant to say was that, we need to take a neutral "take
it from who it comes" approach.

In other words, if the US Embassy has a warning, we can relay that
warning without judging or aligning ourselves with the message itself.
Just the fact that it is an official warning message issued by an
internationally recognised government body is enough.

Mr. Evoltech

unread,
Jan 16, 2012, 8:39:31 PM1/16/12
to swift...@googlegroups.com
I would like to second that this is not a simple question. I have been
working with groups to provide resources to political resistance
movements and actions. We are working on using Ushahidi/Swiftriver as a
platform for aggregating, collecting, and dispatching information. The
scenarios that happen in the streets are often very violent for
participants (violence from police directed against protestors). In a
number of these scenarios twitter accounts that were relaying
information were shut down. In this situation we have found that we can
not rely on twitter.

I am personally very critical of the work done by some large emergency
response groups here in the US, specifically FEMA which was able to help
in some ways also stripped many people of their own agency in being able
to self organize and provide for themselves better then FEMA would.
Governments in general seek to "manage" disaster response and "manage"
individuals. One of the benefits of deployments of Ushahidid and
Swiftriver is that there is a possibility of providing good information
to a lot of people so that they can make decisions for themselves and
self organise. Messages, in any form, are inherently biased by the
author. I think Nick brings up a good point here and that information
coming from social media, depends on the source.

Frank Ostermann (JRC-VGI-FF)

unread,
Jan 31, 2012, 3:24:22 AM1/31/12
to SwiftRiver
While the credibility of social media information certainly depends to
a
large extent on the source, humans use also another heuristic to
judge
the credibility of new information: What do I already know about the
situation and the area? That's why we (a small research team at the
European Commission's Joint Research Center) try to tap into the
knowledge that is already there in the form of authoritative and
volunteered spatial data infrastructures, as well as clustering
social
media info not only for content, but also in space and time. We argue
that by giving social media information back its (geographic)
context,
we can judge credibility and relevance of individual UGC much more
reliably.

Our work shows that you cannot rely well on a single piece of social
media information (unless it comes from an authoritative and verified
source), yet the combination with other information makes it possible
to
automate quality assessment for a large part of the social media
content
on emergencies.

Frank.

Jon Gosier

unread,
Jan 31, 2012, 10:29:46 AM1/31/12
to swift...@googlegroups.com
Great points everyone, but this isn't exactly a new problem. "...information coming from social media, depends on the source." Coud easily read "The reliability information coming from anywhere, has always depended on the source."

Even verified and trusted human sources aren't completely reliable nor are the humans vetting them.  Those traditional information sources are subject to human error, bias, and mis-judgment. The only thing that's changed with technology is that we can capture messages in bulk and somewhat automate the verification process at scale.  However, it's not really a new issue, systems like the ones many people on this list are building simply mimic what humans would do otherwise, just faster and more efficiently. 

Still, I don't really see how this relates to the dissemination of information already vetted and deemed credible by humans back through the channels from which it was captured. For example: 

What's the difference between posting a targeted Hurricane warning on Twitter that was reported by a team of the top meteorologists in the world in a 24-page report, versus posting the same warning when the only report you got was a 140-character message?  The point either way is to warn those in the general public who might otherwise be caught off guard while not bothering the rest.

Is the source less credible because the method of delivery has changed? If so, what are the acceptable media channels: documents? word of mouth? telegram? carrier pigeon?  :-P

--
Jonathan D. Gosier // TED Senior Fellow

skype | j.gosier




Reply all
Reply to author
Forward
0 new messages