This year Parliament hackathon, Accountability Hack, was hosted by Tracy Green (Parliament), Nick Halliday (NAO), Matt Jukes (ONS) and Julia Higginbottom (Rewired State) at the National Audit Office.

dxw team of four developers/designers (Lily, Duncan, Tom and myself) together with many other keen “hackers” turned up at the NAO on Saturday morning to work on the challenges and create something awesome.

Our idea was inspired by the forthcoming HousingDay event on the 12th November. We decided to create an app which will allow to compare homelessness, social housing stocks and social housing that is sold privately in each Local Authority area.

Right to Buy-Bye

The data was mostly taken from the ONS census for 2011 via ONS OpenAPI. The other datasets we used,Live tables on homelessness and Live tables on social housing sales, didn’t have the 2011 data so we had to take the data from the first half of 2012 in some places. Additionally, some information from local authorities was just missing. That’s why we couldn’t draw any conclusions from the comparisons in our project but wanted to give an example of the kind of analysis that can be done if there is reliable and up to date information from local authorities.

Hacking

We built the winning Right to Buy-Bye app using Ruby on Rails, Zurb Foundation, SASS and HAML.

The beginning was challenging because we had some problems getting the census data out of the API. Luckily Sam Machin helped us pulling it from the dataset and we were able to start filling up the empty repo.

You can read about our journey through Accountability Hack on #AccHack Live Blog that was written by Lily.

And the winner is…

We submitted the app on a dedicated page on Rewired State on Sunday and were ready for the 3pm show and tell.

Right to Buy-Bye screenshot

There were five main prize categories: Parliament Challenge, ONS Challenge, NAO Challenge, Multi Org Challenge and Best in Show, plus a few honourable mentions.

The judges, Lee Bridges, Meri Williams, Tom Loosemore, had a tough decision to make to select the best projects. All of the hackers have done great work during the weekend.

Parli-N-Grams by Giuseppe Sollazzo, which analyses the parliamentary debates, won the Parliament Challenge. PFI Explorer by Richard Palmer, Vica Rogers, Xavier Riley and James Southern got the prize in NAO Challenge category. It aims to visualise and explore Private Finance Initiative Data. Edward Wills, Steve Upton, Jake Mulley and James Billingham won the Multi Org Challenge for Politician Personality Test. It’s an app that analyses MPs from their Twitter feeds and Facebook pages.

We were chuffed to hear that our Right to Buy-Bye app won the ONS Challenge and the Best in Show prizes! ONS will be promoting the best developer app at the launch event of their Data Explorer and OpenAPI at BT Towers on 12 November. We are very pleased we will be able to contribute to it.

The organisers did an amazing job at #AccHack14. Thank you for an excellent weekend!

Tweets and feedback

ONS winner @thedxw telling a story to resonate with society with Right to Buy-bye design finesse #AccHack14Tracy Green (@greentrac) November 9, 2014

Really pleased that @lily_dart & @thedxw team won the @ONS category at #AccHack14 – really great app in such a short time. Matt Jukes (@jukesie) November 9, 2014

Very elegant overall winner from #AccHack14 Right to Buy-Bye http://t.co/5CTSlNb6uw with lots of key housing info. #HousingDay Nick M Halliday (@nickmhalliday) November 10, 2014

6 comments

  1. Comment by Tim Blackwell posted

    This is a neat piece of work, demonstrating what can be achieved by a talented and committed team in a very short space of time. Unfortunately, it also illustrates the risks inherent in such a quick and superficial approach.

    The information presented (disclaimer notwithstanding) is *hugely* misleading. To give one example, apparently Camden has (or had) 99.88% of the social housing it needs, with a shortfall of only 38 lets. Try telling that to the families in Camden’s hostels (see http://www.theguardian.com/news/2014/nov/11/-sp-no-exit-britains-housing-trap ).

    It looks to me that you’ve defined satisfaction of social housing need as:

    100*NL/(NH+NL)

    Where NL is the number of social lets, and NH is a count of homeless households falling within a priority needs group, accepted as not being unintentionally homeless and thus owed a main homelessness duty by the authority, possibly the number accepted as such over a quarter.

    This represents an extraordinarily narrow definition of social housing need. It doesn’t reflect the 650 households accepted as homeless by Camden and stuck in temporary accommodation, never mind the 20K or so households on the waiting list for social housing – approximately half of whom had a recognised housing need.

    http://www.camdendata.info/AddDocuments1/Draft%20Evidence%20Base%20Document%20Jan%202011%20-%20version%209.pdf

  2. Comment by Tim Blackwell posted

    This is a neat piece of work, demonstrating what can be achieved by a talented and committed team in a very short space of time. Unfortunately, it also illustrates the risks inherent in such a quick and superficial approach.

    The information presented (disclaimer notwithstanding) is *hugely* misleading. To give one example, apparently Camden has (or had) 99.88% of the social housing it needs, with a shortfall of only 38 lets. Try telling that to the families in Camden’s hostels (see http://www.theguardian.com/news/2014/nov/11/-sp-no-exit-britains-housing-trap ).

    It looks to me that you’ve defined satisfaction of social housing need as:

    100*NL/(NH+NL)

    Where NL is the number of social lets, and NH is a count of homeless households falling within a priority needs group, accepted as not being unintentionally homeless and thus owed a main homelessness duty by the authority, possibly the number accepted as such over a quarter.

    This represents an extraordinarily narrow definition of social housing need. It doesn’t reflect the 650 households accepted as homeless by Camden and stuck in temporary accommodation, never mind the 20K or so households on the waiting list for social housing – approximately half of whom had a recognised housing need.

    http://www.camdendata.info/AddDocuments1/Draft%20Evidence%20Base%20Document%20Jan%202011%20-%20version%209.pdf

  3. dxw staff member Comment by Duncan Stuart posted

    Hi Tim

    Thanks for your comments. I absolutely agree: we made some extremely hefty assumptions and if you look closely at the numbers then in some cases they’re not going to be as accurate as they could be, and in other cases they’ll be downright misleading.

    However, this is missing the point: the aim of this sort of a hack is to demonstrate what *could* be done and to highlight issues with the data (and our understanding of it) which stand in the way of producing an accurate and robust application.

    Although our definition of social housing need is indeed woefully inaccurate, we’re confident that data exists (though possibly not open or consistent data) which would allow us to calculate a much more meaningful measure.

    I’ve written a post which goes into the data we used in detail – including its limitations, our assumptions and compromises, and how we’d improve the app with better data and better understanding: https://www.dxw.com/2014/11/accountability-hack-2014-a-closer-look-at-the-data-we-used/

    And as an open source project you can of course see exactly the calculations we used – warts and all: https://github.com/dxw/AccHack14/blob/master/app/models/local_authority.rb

    All the best

    Duncan

  4. dxw staff member Comment by Duncan Stuart posted

    Hi Tim

    Thanks for your comments. I absolutely agree: we made some extremely hefty assumptions and if you look closely at the numbers then in some cases they’re not going to be as accurate as they could be, and in other cases they’ll be downright misleading.

    However, this is missing the point: the aim of this sort of a hack is to demonstrate what *could* be done and to highlight issues with the data (and our understanding of it) which stand in the way of producing an accurate and robust application.

    Although our definition of social housing need is indeed woefully inaccurate, we’re confident that data exists (though possibly not open or consistent data) which would allow us to calculate a much more meaningful measure.

    I’ve written a post which goes into the data we used in detail – including its limitations, our assumptions and compromises, and how we’d improve the app with better data and better understanding: https://www.dxw.com/2014/11/accountability-hack-2014-a-closer-look-at-the-data-we-used/

    And as an open source project you can of course see exactly the calculations we used – warts and all: https://github.com/dxw/AccHack14/blob/master/app/models/local_authority.rb

    All the best

    Duncan

  5. Comment by Tim Blackwell posted

    Thanks for this considered response – and for your informative further blog post. I fully understand that the application is a showcase of what might be achieved by combining public data sets, rather than purporting to reflect any underlying reality.

    However, I’m not totally confident that this will be clear to everyone -see Nick Halliday’s tweet referencing ‘lots of key housing info’, above.

    Given the above, it’s probably fortunate that the calculated figures in particular are *so* implausible.

    kind regards,

    Tim Blackwell.

  6. Comment by Tim Blackwell posted

    Thanks for this considered response – and for your informative further blog post. I fully understand that the application is a showcase of what might be achieved by combining public data sets, rather than purporting to reflect any underlying reality.

    However, I’m not totally confident that this will be clear to everyone -see Nick Halliday’s tweet referencing ‘lots of key housing info’, above.

    Given the above, it’s probably fortunate that the calculated figures in particular are *so* implausible.

    kind regards,

    Tim Blackwell.

Comments are closed.