Thursday 23 July, 2020

Portrait of one of the latest arrivals in our garden. Shot — appropriately enough — using the ‘portrait’ setting on the iPhone 11, which kept telling me to “move further away”! As Umberto Eco wisely observed all those years ago, the Mac is a Catholic machine: only one true way to salvation.

Click on the image to see a larger version.

Quote of the Day

“I don’t want to be a hero. I want to teach.”

  • Claudia, a music teacher at a school in Massachusetts, who doesn’t want teachers to be put at risk then be lionized, as has been the case for healthcare workers.

Fawkes News: Image “Cloaking” for Personal Privacy

This is lovely. And it’s a student project too.

How do we protect ourselves against unauthorized third parties building facial recognition models to recognize us wherever we may go? Regulations can and will help restrict usage of machine learning by public companies, but will have negligible impact on private organizations, individuals, or even other nation states with similar goals.

The SAND Lab at University of Chicago has developed Fawkes1, an algorithm and software tool (running locally on your computer) that gives individuals the ability to limit how their own images can be used to track them. At a high level, Fawkes takes your personal images, and makes tiny, pixel-level changes to them that are invisible to the human eye, in a process we call image cloaking. You can then use these “cloaked” photos as you normally would, sharing them on social media, sending them to friends, printing them or displaying them on digital devices, the same way you would any other photo. The difference, however, is that if and when someone tries to use these photos to build a facial recognition model, “cloaked” images will teach the model an highly distorted version of what makes you look like you. The cloak effect is not easily detectable, and will not cause errors in model training. However, when someone tries to identify you using an unaltered image of you (e.g. a photo taken in public), and tries to identify you, they will fail.

Fawkes has been tested extensively and proven effective in a variety of environments, and shows 100% effectiveness against state of the art facial recognition models (Microsoft Azure Face API, Amazon Rekognition, and Face++).

They’ve put together a terrific 12-minute video which explains the system. Well worth watching.


Given that facial-recognition is such a toxic technology, it’s great to see its weaknesses being turned against it.

Thanks to Cory Doctorow for alerting me to it.

Where will everyone go?

I hate to say this, but compared with the upcoming climate crisis, the Coronavirus disruption is small beer. Nearly a decade ago the UK Cabinet office did a simulation study to try and figure out the effects of climate change on migration. The results, from my hazy memory, ran something like this. The model divided the global population into three groups: those who live in regions that will be relatively unaffected; those who were too poor to move, no matter how hot or inhospitable their locations became; and those who are able to move when things start to become intolerable where they are. The thing that most struck me about the scenarios was that most migrants will head for cities; and most of their target destinations are coastal cities which are at risk from sea-level rise.

But that was a crude modelling exercise. Now ProPublica and The New York Times Magazine, with support from the Pulitzer Center, have for the first time modeled how climate refugees might move across international borders. Today they publish their findings.

It’s a long, riveting and in some ways alarming read. And a story that’s beautifully told, making great use of the Web and photography as well as supercomputer modelling.

The modellers simulate five scenarios for climate-driven population movement in Southern and Central America.

1 An optimistic/reference scenario, in which climate impacts are rapidly reduced on a global scale and there is regional convergence toward higher levels of development across Central America and Mexico.

2 A pessimistic scenario, in which climate change impacts are on the high end of current plausible scenarios and significant challenges to socioeconomic development exist throughout the region, exacerbating the gap between Central America and the United States.

3 A more climate-friendly scenario, which pairs a less-extreme climate outcome with the same challenging socioeconomic future as the pessimistic scenario.

4 A more development-friendly scenario, which follows the pessimistic climate future but assumes a more inclusive development pathway in which regional economic growth occurs quickly.

5 A moderate scenario, in which socioeconomic development occurs rapidly throughout the region accompanied by a moderate level of climate change.

It’s an extraordinary piece of reporting and investigation. Read it if you can.

Escalation by Tweet

The department of War Studies at King’s College London has just produced an interesting report on the risks of conducting international diplomacy via Twitter, especially during crises.

The Executive Summary reads:

Social media has quickly become part of the geopolitical landscape, and international leaders and officials are increasingly taking to Twitter during crises. For US decision-makers, however, Twitter presents a bit of a paradox: on the one hand, tweets from government officials may help shape the American public narrative and provide greater insights into US decision-making to reduce misperception by foreign actors. On the other hand, tweets may increase misperception and sow confusion during crises, creating escalation incentives for an adversary.

To reconcile this paradox, we examine the useof Twitter by international leaders during crises in recent years, some of which involved nuclear-armed states. In so doing, we explore the changing nature of escalation, which now resembles a complex web more than a ladder, and examine specific escalation pathways involving social media.

Based on this analysis, we find that social media has the potential to be a disruptive technology and exacerbate tensions during crises. To reduce the risk of tweets contributing to escalation in a crisis, we recommend the US Department of Defense:

• lead an interagency effort to develop best practices on the use of social media during crises;

• encourage leaders and officials to refrain from tweeting during crises and instead rely on more traditional means of communication, such as press releases and official statements;

• explore how to build public resilience to disinformation campaigns and provocations via social media during crises, as the American public is asymmetrically vulnerable to these attacks; and

• improve understanding of how various international actors use social media.

Twitter, as a company, and alliances such as NATO, also have a role to play in limiting the negative impact of Twitter during crises. If these findings could be summarised in 280 characters or less, it would be: ‘To manage escalation during crises, stop tweeting.’

One of my favourite accounts on Twitter is @RealPressSecBot — a bot that takes every one of Trump’s tweets and immediately reformats them as an official White House press statement. Which, in effect, is what they are.

This blog is also available as a daily email. If you think this might suit you better, why not subscribe? One email a day, delivered to your inbox at 7am UK time. It’s free, and there’s a one-click unsubscribe if you decide that your inbox is full enough already!