THE BRITISH LIBRARY

The Newsroom blog

News about yesterday's news, and where news may be going

Introduction

Whether you are studying history, politics, society, international relations, economics, media history, sports history or family history, our collections will have something for you Read more

30 March 2017

St. Pancras Intelligencer no. 40 - fake news special

Fake news is probably as old as news itself. Certainly, as far as the British Library is concerned, it goes back to 1614 at least, when the good people of Horsham in Sussex were told of the dragon in their area that was causing great annoyance. Whether those who produced this newsbook believed what they were telling to be "true and wonderfull", who can say? 

Trueandwonderful

True and Wonderfull. A discourse relating a strange and monstrous serpent (or dragon) lately discovered, and yet living in Sussex, 1614 newsbook

Today, the subject of fake news is hot news, coming out of the 2016 US presidential election, but with deeper roots in the clash between traditional news providers and the search engines and social media sites through which so many now discover the news that they want to see. Fake news ranges from deliberate falsity, to news you disagree with, to satire. This special edition of the St Pancras Intelligencer rounds up some of what is being said and done about fake news today.

Definitions

Fake news: what is it, and how can we tackle it? (Digital Social Innovation) - A handy summary from Toby Baker of NESTA

Fake news. It's complicated (First Draft News) - Claire Wardle attempts to explain and categorise the many types of 'fake news'

Lists

Fake News Watch - Want to know what is a fake site, a satire site, or a clickbait site? Fake News Watch attempts to list them (mostly if not all American). Other lists of fake news sites have been produced by ThoughtCo, Snopes, The Independent, and of course Wikipedia

The Ultimate 'Fake News' List (Infowars) - But just to show that one person's truth is another person's outrageous lie, here's an American far right show's listing of the fakery it sees in the mainstream media

Fact checking

CrossCheck - A fact-checking site from First Draft News, formed through a coalition of 37 publishers, mostly from France and Britain, including the BBC, Channel 4 News, Le Monde, BuzzFeed, and Agence France-Presse. Digiday's report European newsrooms are forming a united front against fake news gives the background. 

The Independent is launching a section called In Fact to debunk fake news (The Drum) - The Independent is launching a new section called 'In Fact' in April which will 'debunk spurious stories'.  Other fact checking sites that have popped up include FactCheck, Politifact and Fake News Checker.

Fact Check blog - Channel 4 News has produced a fact check blog following a season of programmes on fake news (including a one-off comedy show). Awkwardly the news programme made a bad slip on the day of the Westminster attack of 22 March, naming the wrong person as the perpetrator, as Richard Smallbrook covers in Westminster attack: Channel 4 learn hard lessons about the fog of breaking news (The Conversation)

Bellingcat Wants Your Help to Debunk Fake News (Vice) - The fact-checking citizen journalism network and scourge of Russia news outlets Bellingcat has launched a Kickstarter campaign to expand its open source investigation platform

RT separates facts from fakes with new online project (RT) - Not to be outdone, RT (Russia Today) has launched its own fact checking service in the battle against fake news, Fakecheck

Facebook and Google

Building Global Community (Facebook) - Mark Zuckerberg has issued a manifesto, which in part addresses the topic of the distribution of fake news (Facebook having been the target of many of the complaints made):

We've made progress fighting hoaxes the way we fight spam, but we have more work to do. We are proceeding carefully because there is not always a clear line between hoaxes, satire and opinion. In a free society, it's important that people have the power to share their opinion, even if others think they're wrong. Our approach will focus less on banning misinformation, and more on surfacing additional perspectives and information, including that fact checkers dispute an item's accuracy.

How Mark Zuckerberg could really fix journalism (Columbia Journalism Review) - Emily Bell responds to Zuckerberg, suggesting that market intervention in America is the answer:

America needs a radical new market intervention similar to that made by the UK Government in 1922 when it issued a Royal Charter and established the BBC ... If, instead of scrapping over news initiatives, the four or five leading technology companies could donate $1 billion in endowment each for a new type of engine for independent journalism, it would be more significant a contribution than a thousand scattered initiatives put together.

Facebook has started to flag fake news stories (Recode) - Meanwhile, Facebook has introduced a 'disputed' tag

Google purges nearly 200 websites in fake news crackdown (Mashable) - Google has been shutting down fake news sites from its advertising platform 

Google's fake news Snippets (BBC) - Rory Cellan-Jones's sneak preview of the Google Home speaker showed how it could spout false news in response to spoken enquiries. Google is now adjusting the algorithms...

Google and Facebook Can’t Just Make Fake News Disappear (Back Channel) - Danah Boyd thinks the problem with the interpretation of news lies with us

Real fake news

How fake news becoming a popular, trending topic (CBS News) - CBS News looks into actual fake news stories created by con artists

Inside the Macedonian fake-news complex (Wired) - More on the production of actual fake news from the unlikely source of the town of Veles in Macedonia

Realfakenews

US spoof news site The Real Fake News

Legislation

'Fake news' inquiry (Parliament.uk) - The Culture, Sport and Media Committee is conducting an inquiry into fake news and its impact

What to know about Germany’s fake-news crackdown (Digiday) - Germany has proposed a law to fine social networks up to €50 million if they fail to remove harmful fake news or defamatory content

More action

World wide web creator Tim Berners-Lee targets fake news (BBC) - Sir Tim has set out a five-year strategy amid concerns he has about how the web is being used

Announcing New Research: "A Field Guide to Fake News" (First Draft News) - First Draft News have also announced a project that aims "to catalyze collaborations between leading digital media researchers, data journalists and civil society groups in order to map the issue and phenomenon of fake news in US and European politics"

Updates from the fake-news world (NiemanLab) - US journalism studies site NiemanLab provides useful round-ups of the efforts being made to tackle fake news. The latest update, Is it still fake news if it makes you feel good?, has interesting points to make about the sharing of positive but made-up news

Historical

Lessons from the fake news pandemic of 1942 (Politco) - There's nothing new under the sun - Joshua Zeitz reports on a race-related fake news story that circulated in the American south in 1942

Trump’s “fake news” playbook has roots in a 180-year-old hoax (Quartz) - Corinne Purtill takes the issue back further to 1835, and the widespread report on life having been discovered on the Moon

The real story of 'fake news' (Merriam Webster) - The American dictionary traces use of the term 'fake news' back to the 1890s - but 'false news' goes back to the 16th century

Opinion

Good news in an era of fake news: the public is becoming wiser about how the media works (The Conversation) - It's an ill wind ... James Rodgers points out that all of this is greatly improving the public's understanding of how the media works

The term ‘fake news’ isn’t just annoying, it’s a danger to democracy (The Independent) - Sean O'Grady is angry

Fake News : The Greatest Lies Ever Told (TruePublica) - So where are the UK's homegrown fake news sites? In a contentious thought piece, Graham Venbergen argues that "In Britain at least, fake news websites have failed to get a grip in the political arena. This is because traditional British news outlets, are already highly accomplished at stretching the truth to its limits and yet still get away with it"

Britain Has No Fake News Industry Because Our Partisan Newspapers Already Do That Job (Buzzfeed) - Jim Waterson similarly argues that very limited appetite for completely fake news in British politics, thanks to its highly partisan newspapers

The Choose-Your-Own-News Adventure (New York Times) - Jim Rutenberg illustrates how we can escape reality by pursuing news worlds that match our expectations. But isn't this how news has always worked?

And finally

'Fake news' to be delightful and fun (Daily Mash) - Let's leave the last word to our favourite UK spoof news site:

The Institute for Studies has shown that real news is bad enough already, and therefore all fake news from now on must be unbelievably delightful. Professor Henry Brubaker said: “If the ‘news’ on social media is just whatever b------- anyone shares, then instead of ‘Muslims in council-backed halal Easter outrage’ why not ‘Puppies discover limitless cold fusion energy source’?

  Cute

http://www.thedailymash.co.uk

03 March 2017

Rand Daily Mail

The latest addition to the electronic newspaper resources available to British Library readers is one that we're particularly pleased to have secured, the Rand Daily Mail. Published from 1902 to 1985, the South African daily newspaper was renowned for its anti-Apartheid stance, with notable coverage of the Sharpeville massacre, the Soweto uprising and the death of Steve Biko. Closed down in controversial circumstances in 1985, the entire newspaper is being digitised and made available by research materials service Readex. Happily the British Library is making the entire archive available for remote access to anyone with a Reader's Pass.

Randdaily

The Rand Daily Mail's renowned African Affairs Reporter, Benjamin Pogrund, wrote recently on the Readex blog about the significance of the newspaper and its archive:

The Rand Daily Mail was ahead of its time in reporting and exposing apartheid evils and in opposing oppressive government. This is why it was shut down. 

The Mail was always a contradictory newspaper: although owned by mining interests from its start in 1902, it was known for siding with the underdog – which, for the first two-thirds of its existence, meant the white underdog. 

That changed in 1957 when Laurence Gandar—a quiet, reserved man—became editor. Little was expected from him except professional journalism. But he proved to have radical ideas and compassion, and he had an inner core of steel. Gandar dissected apartheid with deep and brilliant writing that electrified the country. 

Gandar took his pioneering into the news columns, assembling a staff of journalists whose political views stretched from left to right but who shared a commitment to fair and honest reporting, investigation and robust comment. The newspaper became the pacesetter in illuminating dark corners of South Africa and gave hope to blacks by pointing to a new direction for the country. It transformed itself, the rest of the Press and deeply influenced the political scene.  

The board of (white) directors soon turned against Gandar and in time got rid of him. His successor, Raymond Louw, made his own singular contribution: he invested the Mail with a tough news sense while retaining its policy strength. 

 Integral to this was that the Mail turned the newspaper adage, “When in doubt, leave out,” on its head. Instead, as the authoritarian government’s restrictions grew on free publication, the newspaper sought to get as much into the open as possible. It wasn't always consistent; but right up to the end, even when tight laws and controls were throttling the Press, the Mail ensured that no-one would ever be able to say that they had not known about the ravages of Afrikaner Nationalist rule.  

The Mail was admired by most South Africans of all colors and was honored by its international peers. The reason for respect was why it was loathed by many, but by no means all, within the white community, and they finally prevailed in getting the commercial owners to close it in 1985.  

It's exciting to know that with digitization the Rand Daily Mail's treasure store of information about crucial years in the old South Africa will now be more widely available.

Pressures which led to the newspaper's board seeking a change of policy to reach out to more to a prosperous white audience ultimately proved damaging to sales and led to the newspaper's closure. The current owners, Times Media Group, decided in 2014 to resurrect the newspaper as an online archive and, through Readex, sought out the best materials, including the incomplete run of the title held by the British Library. 

We are delighted that not only is the electronic archive now available in our Reading Rooms, but is available to British Library Reader pass holders via our Remote Resources service. It therefore joins the small but significant number of electronic newspaper resources to which we subscribe that we can offer to Library users wherever they might be, so long as they have Reader's Pass (information on obtaining such a Pass is here). Other titles available in this way include:

  • African American Newspapers, 1827-1998 - provides online access to approximately 270 U.S. newspapers chronicling a century and a half of the African American experience. This unique collection features papers from more than 35 states - including many rare and historically significant 19th century titles.
  • Early American Imprints, Series I: Evans 1639-1800 - contains virtually every book, pamphlet and broadside published in America during the 17th and 18th centuries.
  • Early American Newspapers, Series I - reproductions of hundreds of historic newspapers, providing more than one million pages as fully text-searchable facsimile images. 
  • Latin American Newspapers Series 1, 1805-1922 - part of Readex's World Newspaper Archive, this database provides access to more than 35 fully searchable Latin American newspapers including key titles from Argentina, Mexico, Chile, Brazil and Peru.
  • World Newspaper Archive: African Newspapers, 1800-1922 - part of Readex's World Newspaper Archive. African Newspapers includes over 30 fully searchable African newspapers including key publications from Ghana, Kenya, Mozambique, Nigeria, Sierra Leone, South Africa and Zimbabwe.

We therefore have a particular strong remote access offering for researchers of African news. The Rand Daily Mail online archive is not complete as yet - currently it runs 1937-1985, but eventually it will contain the full run of 1902-1985.

11 January 2017

Analysing the past

There are exciting changes happening in how we use newspapers to study the past. After decades in which the use of newspapers in research meant leafing through volumes or scrolling through microfilms, digitisation made millions of newspapers more readily searchable and far more widely available. But now that digitisation that taken us to the next stage in development, which is using the data generated by the digitisation process to look at history on a grand scale. We are moving into the era of big data newspaper studies.

Peopleinhistory

From the University of Bristol study: People in history. (A)  famous personalities by occupation using all extracted entities associated with a Wikipedia entry; (B)  the probability that a given reference to a person is to a male or a female person

Big data newspaper studies have come about through a combination of large-scale digital resources and a growth in analysis tools. Most will be aware of OCR (optical character recognition), the mechanism by which archival texts can be converted into machine-readable texts by converting what a computer sees as an image (i.e. the arrangement of letters on a page) and matches these to letters that it knows. It is an imperfect science, because OCR can struggle to work with older forms of types and deteriorating page originals, but levels of accuracy continue to improve as new OCR software is developed, and the results are generally satisfactory - that is, most of the time a researcher will find what they are looking for, if it is there to be found.

But added to this are software tools that can extract further sense from the raw data set that generated by OCR. The field of what is called Natural Language Processing, by which computer come to understand human text and speech, includes the extraction of keywords, or named entities, and the matching of these to controlled lists of terms (such as DBpedia), further mapped to geographic areas and time periods, which enables researchers to undertake controlled, thematic analysis of large historical datasets. Our archive of words yields patterns of behaviour with much to tell about our past selves.

This is the theme of a major project undertaken by the Intelligent Systems Laboratory at the university of Bristol, led by Professor Nello Cristianini. As described in their paper 'Content analysis of 150 years of British periodicals', the project worked on a corpus of newspapers digitised from the British Library's collection by family history company Findmypast for the British Newspaper Archive website. The figures involved are huge. The project analysed 28.6 billion words from 35.9 million articles contained in 120 UK regional newspapers over the period 1800-1950, which they calculate forms 14% or all regional newspapers published in the UK over the period.

The project then used this study to explore changes in culture and society, determined by changes in the language. It looks at changes in values, political interests, the rise of 'Britishness' as a concept, the spread of technological innovations, the adoption of new communications technologies (the telegraph, telephone, radio, television etc), changing discussion of the economy, and social changes such as mentions of men and women, the growth in human interest news and the rising importance of popular culture. It is the stuff of multi-volume histories of the past, boiled down to eye-catching graphs.

This does not mean that we thrown away those multi-volume histories, however, The researchers are at pains to point out that such data analysis is an inexact science, with many caveats needed to explain how the entities have been arrived at and with what degree of caution they should be treated. The data derived from such tools can only work where it is supported by traditional studies, to gain the richer understanding of what happened. The machines may have taken the natural language of humans and converted it into data, but the results need to be converted back into human language to offer real understanding.

So it is that some of the results of the project yield results that may seem obvious. We could have guessed beforehand that the newspaper archive would show an increase in discussion of popular culture subjects, that politicians are more likely to achieve notoriety within their lifetimes than scientists, or that there was a rise in coverage of the Labour Party from the 1920s onwards. But the analyses reinforce through data what we have previously inferred through study, while discoveries such as the term 'British' overtaking the term 'English' at the end of the 19th century, or the decline in terms associated with ''Victorian values - such as 'duty', 'courage' and 'endurance' - call for new studies to explore these things further.

The project is at pains to point out the importance of using newspaper archives. Previously we have had big data analyses of millions of historical books, most familiar through the Google Ngram Viewer. This has caused controversy among some scholars, because of the unevenness of coverage of topics in books, and the limitations of merely counting words and making them searchable again. Opening up newspaper archives for comparable analysis widens the amount of content available, arguably with greater reliability overall, and now with tools to make analysis that much more scientific. The use of controlled terms will also enable the analysis across different datasets - so, books and newspapers, but also other news forms, as subtitle extraction and speech-to-text technologies now start to make our television and radio archives available for similar and shared analytical studies. Our big data is only going to get bigger.

There are limitations to this use of newspaper archives. The quality of OCR varies not only according to the original newspaper, but according to the microfilm where this has been used instead of print. Digitisation is quicker and cheaper this way than digitising from print, but older microfilm can be photographically poor, leading to inferior OCR (though there are promising tools appearing for improving poor OCR). The British Newspaper Archive is made up mostly of UK regional newspapers, because the main nationals have often been digitised by their current owners and are available separately. How different was the discourse in newspapers based in London from those around the rest of the country? That has to be the subject of another major study.

Meme

One of the better jokes from the Victorian Meme Machine project

The British Library has been engaged in its own big data analyses of newspaper archives. BL Labs is an initiative designed to support and inspire the public use of the British Library’s digital collections and data in exciting and innovative ways. It has facilitated several studies of British historical topics through the digital newspaper archive. These include Bob Nicholson of Edge Hill University's study of jokes in Victorian newspapers, with the concept of the Victorian Meme Machine (automatically matching jokes to an archive of contemporary images); Katrina Navickas of the University of Hertfordshire's mapping of nineteenth century protest; and Hannah-Rose Murray of University of Nottingham's tracing of black abolitionists in 19th century Britain. A major user of our newspaper data is M.H. Beals of Loughborough University, who is researching how ideas travel across the historical news media, creating new insights through understanding newspaper archives as structured data.

Such projects are just the start. The availability of large-scale newspaper archives in digital form, and the data derived from such archives, enables us both to seek answers to traditional questions more quickly, and to start asking new kinds of questions. The latter is the great challenge that newspaper data offers. We need to come up with new questions, because the technology enables us to do so, and because it may question what we previously thought that we knew. As the data from their archives comes more readily available, and more easily usable by the non-data specialist, so we will find that we have only just started to read the newspapers. We are going to find that they have much more yet to tell us.

Links: