Digital threats to democracy and what to do about it: Speech at the 2018 Re-Publica Media Convention Berlin

Liz Carolan at the 2018 Re:Publica Media Convention Berlin, May 3rd 2018

Below is the text of the speech I gave at the 2018 Re:Publica Media Convention in Berlin, on May3rd 2018. It was 3 weeks before the referendum and I was giving an outline of what we have found in the Transparent Referendum Initiative project, and what it meant for policy discussion.

I’m here today to talk about a live example of a digital threat to democracy that is happening right now Ireland.

In 3 weeks we will have a referendum on whether to maintain or remove the near total ban on abortion — it is a vote that is highly emotive, contentious, and its of global interest and symbolic relevance to people inside and outside of the country.

I run a project that is capturing and allowing real time transparency of social media ads during this referendum.

With this project we have been able to build a picture of what is going on, as it is happening — an incomplete picture, but a picture none the less.

I believe this is the first time that such information has been available to the public in real time during a referendum or election.

There are a lot of things we have already learned since we started at Christmas.

Back then we had a hunch that we might uncover some examples of international attempts at influence — and indeed we did. But it looks quite different than what we thought.

Back then we thought that we might see some activity that breaks Irish rules — we didn’t realised then that we would uncover new ways that people are using social media to undermine those very rules.

And back then we didn’t realise how just powerless the institutions we have would be in protecting or at a minimum responding to these threats to our democracy.

As I said we’re still right in the middle of this so I really do welcome the opportunity at this crucial stage to take a step back and share with you — for the first time — what we are seeing.

And in particular I welcome the opportunity to share the stage today with a regulator, a politician and representative of Facebook.

I’m going to talk you through the example of Ireland, but only as context so that we can have a broader conversation about what can be done to make sure that the promise of social media and the online world — that it would help rather than hurt democracy — becomes a reality.

Because at the moment, in Ireland, right now, it doesn’t always feel like it is.

2 — CONTEXT

Let me start with some context.

As I said Ireland is having a referendum on whether or not to allow legal abortion.

At the moment we have one of the strictests bans in the developing world — including in cases of rape or where a baby won’t live past birth. This ban is enshrined in the constitution in the so-called “8th amendment”.

Those who support the 8th amendment claim it has saved thousands of lives of those who would have otherwise never been born.

Those who oppose it claim that 12 Irish women per day have abortions anyway — either by traveling to England, or taking unregulated pills purchased on the internet. They also point to a small number of high profile cases where women have died or been denied medical treatment because of the strict rules.

So there is a lot at stake.

I work in the area of transparency and open data, mostly in developing countries, but I am also an election nerds and had been following the developments and questions arising from the use of algorythms and social media ads in politics.

Having been in DC for the Trump election, and in London right after the Brexit vote, I took one look at the referendum about to happen in my own country and thought — we need to make sure we are watching the online space in this one.

3 — PROJECT

So, myself and some friends started this project — the Transparent Referendum Initiative — that aims to bring social media ads into the light so they can be scrutinised.

There were three of us at the beginning, myself, Craig Dwyer — a social media campaigner for the gay marriage referendum in 2015, and Peter Tanam, who runs a social media ads agency — two politically savvy tech optimists.

We identified three potential risks social media advertising could pose in this referendum:

- That outside actors would purchase social media ads targeting people

- That misinformation would be spread in ways that are difficult to detect

- And that large-scale spending would be happening under the radar

Our initial vision for the project was to partner with the platforms to build a public, open database of all political ads.

Twitter told us they were not accepting ads for this referendum, and Facebook said at the time that it would be against the contracts they had signed with advertisiers.

So we partnered with a UK organisation WhoTargetsMe who had developed a chrome extension that collects data about ads people see on Facebook on desktop and laptop.

We had no money, but we chipped in €50 to buy a web domain www.tref.ie and set up some Twitter and Facebook accounts, and asked people to get involved.

About 600 people are now using WhoTargetsMe, and through them we have collected metadata — links, account names, numbers of likes etc. — of almost 500 ads.

We also have been collecting ads that are shared with us via screengrabs, which include ads on facebook we didn’t pick up via the plugin, but also a very large amount of activity across google’s ad services and YouTube.

A core part of our strategy has been to share ads we are gathering with the public, but also with journalists who can do the digging required to turn this data into meaningful stories. This has led to stories everywhere from the BBC, to Wired, the London, Irish and NY Times, Buzzfeed and a host of other publications.

So, here is what we — and they — have seen.

4 — PROXY WAR

For a start, that overseas organisations are buying ads on facebook targeting Irish voters with referendum related content.

However, the examples we are seeing are of small purchases by organisations and indiv — including in the UK, New York State and Canada.

So what is interesting here is that unlike the coordinated and deliberate attempts at interferance by Russia in the US election, this appears to be overseas interferance in a political process by private citizens and organisations, enabled by social media.

We have highlighted this with our regulators, but in Ireland while we ban overseas donations to campaigns, there is no rule against direct expenditure of this sort — in 1997 when the laws were written it wasn’t possible for someone in New York to try influence voters, without an incredible level of sophistication.

Journalists have followed up with the people placing them — one said they were asked by a pro-life group to target Irish people — but they could also just be citizens spending small their own money to promote their ideas.

This would not be a big deal if it were not for the broader context.

So while I mentioned that this referendum is significant within Ireland — it is seen by folks on both sides as a matter of life-and-death importance -

This referendum is also significant outside of Ireland, for different reasons.

Ireland is seen as somewhat of a symbol globally when it comes to battles between forces of progressiveness and forces of conservatism. The 2015 referendum on Gay Marriage led to headlines around the world. It was seen by many as a symbolic victory for progressive causes.

But similarly, Ireland’s constitutional abortion ban is held up by many conservative pro-life groups around the world as an example to be emulated — and as a symbol to be maintained.

In fact, in the run up to this referendum, there have been campaign groups in the US openly fundraising to pay for campaigning — to fund the “culture wars” in Ireland — and there is at least one group of US campaigners arriving this week to an AirBnB they have rented using $8k of crowdfunded money.

I do not think it is an exageration to say that for some people, this referendum is almost a proxy war for a cause they believe in — a soft power proxy war, being fought mainly by non-state actors — by citizens of other countries.

None of this of course is Facebook’s fault, but it is in some ways being enabled by that and other companies that operate across borders.

I was contacted last week with one of the founders of a site that ended up in our database — he claimed that they were using ads not to influence the referendum, but as part of their ongoing global community building work.

We know that online communities are incredibly important spaces for activitists and minnority groups of all persuasions, and that is something that we do not want to lose.

But the question from a policy point of view, is where do we draw the line on this stuff?

On the one hand there is technology enabling and empowering global communities of interest to share ideas and find likeminded people

On the other is the need to maintain the integrity of sovereign democratic processes, because only those who live within a national border will have to live with the consequences of a vote.

Where do we draw the line?

5 — ADS AS WAY TO IDENTIFY OTHER ACTIVITY

Another finding from our project, is that Facebook has been a positive force in enabling identification of more worrysome overseas attempts to influence the vote.

If you are a person setting up a website or trying to get messages to voters, without substantial networks in that contry, Facebook, Google and other platforms give you a data-driven way in.

An example of overseas connected activity just this week, uncovered in part through our work bringing, was of a US pro-life group using Facebook to promote a website doing misinformation — it was posing as neutral source of facts — in fact it was collecting data through cookies which could be used for further ads.

In this case the Facebook ads were the crucial“last mile”for connecting overseas groups to Irish voters and transparency on those ads gave the clue to follow to build the bigger picture.

Just like we know that the Russian governmnet connected Internet Research Agency had a monthly budget of millions of dollars to interfer in the US election — only a small fraction of that went on actual facebook ads — but that is the point where the activity can come to light.

So in this context, we can see the power of transparency of ads on these platforms to allow for broader scrutiny of activity related to the referendum.

Here the policy action is to have full, transparency of all ads — who is placing them, who is being targeted, and for that to be available in real time. This is what we have modeled in our project, so we have shown that it is possible.

6 — TARGETING AND ANONYMOUS PAGES

We thought we would see examples of micro-targeting by the campaigns — delivering different messages to different people.

We have seen some of that in relation to individual ads, but actually what is more interesting in that our data appears to show patterns multiple of anonymous, untraceable accounts targeting different groups with ads.

In our database, which again is only partial, we have found about 160 different pages placing ads, and about another 30 have been flagged with us in screengrabs.

We have seen examples of pages who have popped up and targeted — for example — young women with disturbing images and lies about death rates related to abortion.

We have seen a page which targeted huge numbers of people with a lie about the source of funding for one of the campaigns.

In that case, and in many cases where misinformation is being targeted deliberatly — the proloferation of pages makes it hard to even now these are there — many of the pages have already been deleted, leaving no trace that they were ever there — except for our database.

These pages have some characteristics — they can have small numbers of followers and very little information about who is behind them.

Many of these accounts that we have picked up, have already been deleted.

The result is combined with the complete lack of tracability of ownership of pages — we do not know if they are activities of bigger campaigns that they do not want to put their brand to, we do not know if they are external influence, we do not know anything.

Facebook continues to allow people targeting Irish voters to share misinformation with them without forcing them to reveal any information about their identity.

This is the real issue with micro-targeted misinformation, as far as we can see in Ireland

So one policy solution we are looking at is to mandate the removal of the protection of anonymity where someone is doing any form of political advertising including ads related to social issues.

The second part of this is to have a low threshold for registration of campaign actors in referendums.

In Ireland and many countries we only require organisations in receipt of large amounts of funding to register.

It made sense in a pre-internet era to exclude small organisations from the regulation net — it was burdensome to register when paper and postage stamps were required.

But now that technology can enable unscrupulous groups to split up or to mobilise and deploy as disconnected networks of anonymous accounts — small amounts of money can quickly buy up access to voters in very influential ways.

7 — SPENDING

I just want to quickly mention before I wrap up one of the questions I wish I had an answer too, that journalists keep asking us and we can’t answer — and that is how much money is being spent in this referendum, and where that money comes from.

This is very hard to tell from the data we have, as there is not fixed pricing for ads, and engagement metrics can be dependent on so many factors.

But quantity of money is one of the thing that we care about in democracy — it can disseminsate information beyond the strength or merit of the argument — and paying for reach on social media to end up with lots of “likes” can make an idea look more popular or legitimate than it is.

But also spend on dissapearing or hard to detect ads is hard to keep track of — which makes it easier to spend money that is unaccounted for. Google ads and YouTube are particularly problematic in this area, and have yet to receive adequate scrutiny.

Policy solutions:

So if I were to pull together what we have learnt, from the point of view of policy, I would say what is needed is:

1. A full list of all ads being shown to voters, in real time, from all advertising platforms, together in one public place.
2. Anonymity / untraceability to not be allowed for political ads — even where amounts of spending appear small
3. A discussion about international influence, and some parameters around how we enable global conversations but maintian national legal sovereignty
4. We need to have an idea of how much is being spent online for political purposes

This is what we are modeling with out project. However we are very limited in what we can do, and there need to be laws mandating this.

In Ireland I think this has to come as reform to our Electoral Acts, but I know that our government ministry with responsibility for communications is also looking at it.

Facebook have started to take action, and Google have some plans, but there needs to be a uniformity of the demands across platforms — this will maintain a level of market competitiveness, but also so we do not end up with one unregulated space where bad actors congragate.

Lets fix this for everyone not just ourselves:

Thinking back to this referendum, there is no doubt in my mind that social media and other digital chanels are a Wild West allowing activities not in line with the democratic values that underpin our laws.

We do not know if this activity will have an impacton the results.

Campaigners and commentators I speak with insist that this referendum will be won and lost in conversations between people. I am not as confident as some of my compatriots that we are immune.

But I do want to end by taking a broader look than just Ireland. When I saw the extent to which this referendum was becoming a proxy fight, I thought about many of the countries I have worked around the world which do not have robust democratic cultures and institutions, and where citizens cannot rely on their governments to bring in policies that can protect them from these kinds threats to democracy.

This is why I believe that the companies involved — Facebook but also Google, YouTube, Twitter and others — need to ensure that their products are living up to the visions that their companies were founded on — of enabling rather than threatening democratic engagement, open debate and discussion.

They need to do this in countries where the rules will not force them too, and I think it is our role in Europe and beyond as we puch for these transparencies at home to make sure that we use our power to grant other citizens around the world the same thing.

Thank you.