Datacasts to put powerful under scrutiny

Transparency campaigners have launched a bid to turn the speeches of powerful people into data the public can easily scrutinize.

In the wake of the Conservative Party deleting its public speeches and election pledges, the initiative may put pressure on political parties not only to publish their speeches but to make them machine readable and primed for mining by data scientists.

Campaign group MySociety yesterday unveiled a system to capture public statements and keep them in the public domain, and to bring the past speeches of powerful people back into the sunlight.

MySociety founder Tom Steinberg cited politicians, business executives and law courts as particular targets for a campaign that proposes to make them more accountable to “normal people”.

It plans a SayIt system to monitor parliamentary proceedings in South Africa and has established a partnership with Chilean transparency group Fundación Ciudadano Inteligente.

But it launched SayIt in the UK first – with a call for people to capture public meetings and statements made at local councils, where most of what is said goes unrecorded.

Steinberg told Computer Weekly the system could also address the problem raised by the Conservative Party’s deleted speeches archive.

“The whole business of transcripts – whether its parliament or celebrity interviews – needs bringing up to speed. It’s woefully old-fashioned.

“We want people all over the world to find out when powerful people talk about things that matter to them. If someone is talking about your road in a local government meeting today, you will never know. It’s impossible to say ping me a tweet if someone in a council meeting talks about my child’s school or whatever,” said Steinberg.

Councils historically simply didn’t publish transcripts, he said, for reasons of cost and politics. The meeting minutes they published instead were “deracinated” and lacked humanity. MySociety was attempting to develop a system that would cut the cost of producing transcripts and turning them into a structured, machine readable form. It is exploring the use of speech-to-text software.

Speech data format

It has also adopted an open data standard called Akoma Ntoso (a West African phrase meaning “linked hearts”), and called on others – including the UK parliament – to adopt it too. It has envisaged a world in which all public statements anywhere could be automatically cross-referenced by person, say, or topic, because the transcripts all used the same open format.

“It’s frankly unlikely that any two transcripts anywhere in the world have the same data structure today because each transcription company, and each court and each parliament will just open Word and start typing,” said Steinberg.

“Whereas structured data is what makes each Tweet a beautifully-structured thing. It’s what makes Facebook work. And it’s what makes everything on the internet work. But as of today transcripts are not structured at all,” he said.

The UK Parliament already publishes Hansard transcripts of its daily proceedings, which MySociety turns into a structured format for publication on its TheyWorkForYou website. But data discrepancies require manual intervention on a daily basis, said Steinberg.

It converted transcripts of the UK Leveson Inquiry and the trial for war crimes of Charles Taylor, former president of Liberia, West Africa, to publicise the launch of SayIt.

It took two weeks for a programmer to convert about 400,000 trial transcripts because they contained so many inconsistencies in naming conventions and other data elements.


Shela Husain, deputy director of accountability and transparency at the Department for Communities and Local Government, said it directed councils to publish minutes of public meetings in its Code of Recommended Practice on Transparency. But it would not ask any more of them.

“We certainly won’t be stipulating that they publish transcripts of meetings,” said Husain. She insisted local authorities should be open, transparent and accountable. But transcripts would incur costs and the question of producing them had never come up.

“It’s up to local authorities,” she said. “We wouldn’t say they shouldn’t do it. We just wouldn’t stipulate that they should.”

Councillor Nigel Murphy, cabinet member for digital at Manchester City Council, said: “It’s something we would look into because its a way we could engage and be more accountable. Transcripts can help people with disabilities. But there’s obviously a cost involved.”


Steinberg said he hoped journalists would use the system to datacast transcripts of interviews and conferences. He said they could share the labour of keeping tabs on powerful people.

One of those powerful people – Google chairman Eric Schmidt – had helped the Conservative Party election campaign by backing its claims that the internet would make people like him and them more accountable. But Computer Weekly showed recently how they had let the ball drop when his and the party’s transparency speeches were purged from all but the most obscure corners of the net.