The Day After – Lessons Learned from the Crowdmap experience

Late on Monday night, I wrote a short post here in anticipation of the crowdmap I’d just set up for BBC London, which I hoped would provide a useful service the following day for the London tubestrike, 7th September 2010.

It’s now Wednesday morning, and I can write, while still feeling slightly shell-shocked from the experience, that all in all, I’m very pleased with how it went.

I want to use this post to reflect on some of the things that worked, some of the things that didn’t work as well, and some things I will do differently if the next scheduled tube strike goes ahead.

Bottom line was that lots of people saw it: 18,860 unque visitors, and 39,306 page views from 55 countries. 13,808 were from the UK, 3863 from the US, and I can’t get over the fact that we had 2 people form Bermuda, 1 person from Uruguay, and 9 from Kenya, the home of the Ushahidi platform. The power of social media never ceases to amaze me.

We posted 202 reports yesterday. About 50 were sent directly to the map from the audience, either via the web form or the specific SMS channel we set up. The rest of the reports we took from twitter, either tweets in the #tubestrike stream or replies to the @BBCTravelalert account.

I can’t stress enough that getting the reports up wasn’t easy because of the time pressures. Every report, whether it was sent directly or not had to be physically approved. Nothing went straight up onto the map.

Yesterday I was ably assisted by Abigail Sawyer who works for the World Service and who wanted to see how the platform worked and how it might work in a Global context, and for 2 hours during the evening rush hour by Emma Jenkinson, a producer from BBC London who was drafted in as emergency help. We also had help from Steve Phillips, the BBC London transport reporter who was audiobooing, appearing on TV, and updating twitter like a mad thing.

During the two peak times, we were monitoring the SMS console, three twitter streams (#tubestrike, “tube AND strike”, @BBCTravealert), audioboo, emails and the BBC London facebook page.

For each report we needed to add or check:

1)    a clear headline,

2)    a description, which if it was from twitter we were cutting and pasting,

3)    the official timestamp (which frustratingly never stayed connected to the actual time so drop down menus had to be used each time),

4)    the geo-location by putting in the location box and waiting for the map to find it (we soon learned that if you just put in Waterloo, it defaulted to Waterloo in Canada so we had to write Waterloo, UK),

5)    the category (tube, train, bus etc)

6)    the verification status (we only ticked the verification box if the report had been supplied by our own reporters. We realised we couldn’t even verify information from the Transport for London website as commuters were contacting us and saying the TFL information was not up to date)

Only then could we finally approve it and then put it on the map.

Phew. Quite a process.

If you had an event which wasn’t so time sensitive or fast paced, it wouldn’t have been such an issue, but at times we were mopping sweat off our brows, feeling slightly under pressure, especially as we saw so many people tweeting about us from around the world!

That was the process.

In terms of things we learned along the way…

I) had originally chosen Google Maps as the default mapping tool, but half way through the morning rush hour we heard from Harry Wood who encouraged us to use Open Street Map,  a free editable map of the whole world, created by volunteers. It is not for profit and apparently started in London. We quickly changed the settings with one click and were immediately amazed at the improved quality of the map. It was much more accurate.

2) Although we needed to use the inbuilt time stamp, we also realised people needed to quickly see on the map itself (rather than having to click through) when information had been sent, so at lunchtime we started each headline with a time stamp we typed in.

3) At lunchtime, we had collected 90 reports, but realised it was quickly going out of date. We therefore deleted all the earlier reports and started afresh, although we did manually input all station closures, which we realised was the key bit of information people were looking for. One major problem however was that by early afternoon word had spread and I saw people tweeting – ‘good idea, shame there isn’t more information on the map’, so I was torn between trying to make the map look impressive, and it actually being useful!

Things I wished we could have done:

1)    Publicised it more beforehand. This was a crowdsourcing initiative but we didn’t talk to the crowd early enough to encourage people to take part, and to show then how it might be helpful. For obvious reasons, this was very much an experiment and the BBC was slightly nervous about shouting about something that hadn’t been tried and tested. As a result, I only published my short blog post on Monday night and we started tweeting about it on Tuesday morning but that was it. So the fact that we got the results we did, are pretty amazing (I’d say modestly!)

2)    I wish we could have had more time to thank people and to let people know we’d used their information on the map. I did it a few times when I got a chance, and unsurprisingly we saw those people posting more reports.

Things I’d encourage Ushahidi to think about:

It feels churlish to make suggestions to the platform, when I think it’s amazing and I wouldn’t have the  skills to make 1/100th of the site, but as someone who used it under pressure in this situation, here are a couple of suggestions:

1)    It would be useful if there was a scrolling news bars at the top, so we could put out top line information, which we know everyone would see by just going to the map. Something like ‘the circle line is suspended’ or ‘the roads are really starting to build with traffic’ was very hard to map. There’s no one spot on the circle line (for those who don’t know, it’s an underground line which runs in a continuous circle!)

2)    It would have been great to add more information to the first speech bubble which appeared when you clicked on a dot, e.g. a photo, an audioboo, more detail etc. I don’t think everyone was always clicking through to the next page.

3)    A way to visualise the timestamp more clearly from the map would have been great, e.g. the brighter the colour, the more recent the report. It was a shame to have to delete earlier reports.

4)    A way to differentiate between good and bad news. Most of the information we were reporting was negative – tube line suspended, traffic jams etc. Sometimes we got tips or advice about how to avoid the problems, and it would have been great if we could have shown those in a different way.

Overall, we created a map, which at many points during the day was more accurate than the Transport for London website, and which was a live and updated version of what was happening out on the streets of London. And most importantly it was built by the people of London.

If more people had known about it and understood how to upload reports it would have been even richer and even more useful and accurate.

While I don’t wish another strike on anyone, I secretly hope there’s another one so we can take crowdmap for another test drive.

27 thoughts on “The Day After – Lessons Learned from the Crowdmap experience

  1. Have to say ushahidi did offer us a bespoke version when we were setting this up – but time was a bit of an issue – am sure they would be up for modified version in future. Also think you’d have masses more traffic if publicised on air in advance so worth asking them about heavy usage – something they also offered when I was speaking to them. They rock.

  2. This is a great round up of how things went. Indeed the pressure of timeliness and accuracy is something that usually comes with practice – esp. for online journalists who put forward material through the day. The UGC hub are great at this sort of thing.

    The time scale could be easily solved with a slider…then your pins will appear and disappear as you adjust it. Probably best set to the current time, but nice to describe a picture of a day or even much longer timescales.

    Ushahidi provides so much as a base and an excellent example for visualising dynamic news, I cannot help but feel that news outlets should be the ones responsible for the “news bars” you mention here. I guess like reskinning to suit a particular outlet or theme.

    In that sense – OpenStreetMap (was used by and supported by Crisis Commons and saw excellent use in Haiti where the emergency services were downloading it as the older maps became less useful when there were blocked roads etc) is a great solution – as it is open, it means that there are more options for tailoring the maps to suit.

    Just means having the right coders and machine translators to hand along with verification and data input.

    Nice to see this being used in a real scenario. Great stuff.

  3. had a look at it yesterday from a non-londener-traveller perspective interested in using ushadidi, its true the dots weren’t very informative, were you resending out the reports via the bbcalert twitter or how many signed up to the alerts?

    is there a way to set the timeline so its default shows the last hour, or give the reports a timespan?

    did you use swiftriver?

  4. Claire, this is some of the most valuable feedback we’ve received since launching Crowdmap a couple weeks ago. Thank you very, very much. We’re taking these in and adding the ones we don’t already have into our dev cycle.

  5. (This is a paste-in of my comment over at a paste-in of your assessment concerning the Crowdmap at Paul Bradshaw’s ‘Online Journalism Blog.’)

    As I’ve said a few times when ‘socializing’ online – content is king if you want a diverse readership. However this is a one-off or more accurately an occasional thing which is very acute. This means a low potential for word of mouth – that you attained such a decent amount of visits in light of the scenario must be encouraging.

    Speaking for myself, I couldn’t do this work due to the heavy dependency on time the crowdmap required through updating. Too sedate and slow, like a lumbering alligator. You’re essentially partaking in an engaged version of the prevailing UK press orthodoxy which is to remain indoors much of the time and work on stuff coming in whether it be from the locality or the PA wire.

    Were you to do this with greater efficiency I’d hire a few people to get out there on the ground rather than rely on SMS updates. Have these individuals comb an area and constantly update you via mobile or a tablet or whatever. Since it’s only a few days work I don’t think you’d spend all that much on paying a few competant people to augment a prospective future mapping.

    As you imply…getting the word out is imperitive – why not try to spread word to get people to subscribe via SMS for updates? Store the numbers and broadcast to them at the next strike.

    I’ve pontificated on this blog about the power of the Internet for media, there was a fun discussion on this very blog: The WWTDD effect.

    All your improvement points tend towards the technical – of course there’ll be rough curves to sand. Google took years to attend the aesthetics and technical finesse it today possesses.

    I’m bias because I think journalism has declined in tandem with old fashioned roaming, picture-taking and writing which – before my time – flourished with thousands of freelance non-paparazzo hacks making a living. Scaled back greatly now, in my view to the detriment of quality.

    Though heavily marginalized or doing reportage mainly as a labour of love thing rather than as a living there are still able workers out there who’d jump at the chance to monitor and report something like a Tube strike for a few days.

    Yes you can count on the public and radio traffic reports to an extent; but do consider that your officebound staff may find it prudent to get some people great at speedily relaying info outside working for your outfit – London is huge so I’d divide the areas of concern into zones…assign one person to a zone or something. Just some thoughts.

    Pete, editor at Dirty Garnet.

  6. Thanks all for great feedback. Really useful to hear other thoughts and opinions about the user experience. It sounds like ushahidi are really appreciating the feedback.
    @jamillah – agree that the UGC hub are very good at this. Hoping that if there’s a next time, some of that experience can be used to help resource the initiative.

    @steve – a way of identifying when information has been posted/how old it is will definitely be a priority for any future mapping project.

    @Erik – hopefully we can have a more detailed chat at some point.

    @Peter – thanks for your comments. Interesting idea about a more formalised ‘community reporter’ idea.

    1. Indeed, which was alluded to – there’s always the traffic reporter who caters to radio for instance who can in times of a Tube strike adapt themselves to relaying info on the problems caused back to radio, and perhaps in future online-based efforts to inform such as this one.

      The problem is is that a ‘few’ is not enough. London is a huge place – hence my idea that more folks familiar with the lay of the land, are nimble and tech-savvy, and would happily do the job for nominal recompense; a note on the CV saying ‘Worked for BBC London, I helped with bla bla…’ is far from worthless in itself.

      So yes by all means keep the radio journos, and the public, engaged at the next Tube industrial action…but get some local aspirant hacks in to bolster and augment the efforts. I’m sure if this idea – based on the promising reception in this instance – gets off the ground then funding for paying the costs/expenses surrounding this idea will happily become available.

      To quickly add; a vital point is to dispel the idea that keeping Crowdmap ready in the background like some Green Goddess for the firemen strikes is somehow cynical. Given the long history of Tube strikes it’s an if not when matter; so if this could be formalized within BBC London, marketed to Londoners in say…a month running up to a strike it could be successful rather than a transient notion neglected despite promise.

      Must go as I’ve obtained a large Times crossword book; these books are now one of the finest products of that News Corp rag and a lunchtime is much happier if frittering it away doing giant crosswords than reading the alleged ‘news’.

      Pete, editor at Dirty Garnet.

  7. Pingback: Ear Pitch Training
  8. I totally hear what you’re saying regarding publicising experiments. We’re running into the same issue here at the ABC. And in this particular instance, it would have been awful if people were relying on the info and then the project failed to deliver.

    It also raises the issue of resourcing. Even with minimal exposure, it sounds as though the project almost failed due to its success. It would be good if there was some way of utilising the crowd in the verification/filtering side of the equation, as well as in the supply of raw data.


    1. @dailexpense thanks for this link and your comment. Great template for user instructions. I talked to Erik from Ushahidi today and fed back that crowdmap wasn’t updating tweets quickly enough, an issue he wasn’t aware of. Once that has been sorted, we would definitely be clearer in our twitter instructions. Erik also explained that Swiftriver is pretty much ready to go, so hopefully we’ll be able to integrate it with crowdmap soon.

  9. Greetings from Los angeles! I’m bored to tears at work so I decided to check out your blog on my iphone during lunch break. I love the information you provide here and can’t wait to take a look
    when I get home. I’m surprised at how fast your blog loaded on my phone .. I’m not even using WIFI,
    just 3G .. Anyways, fantastic site!

Comments are closed.