By Nanistya Probosutedjo, Data Analyst at Zero One Technology
Welcome to Zero One Group’s very first newsletter report!
This newsletter started in an effort to “Think Better, Not Perfect” and keep learning. Hence, we decided to introduce the newsletter to keep up with the latest events and learn collectively. What makes our newsletter different is that we try to select topics with actionable items. We’ll get back to this later. If you are in the tech industry or aspire to join one day, we hope this newsletter will inspire you to explore new corners of knowledge with us!
The topic for today’s newsletter will focus on research-based COVID-19 topics and self-help apps to get us through this difficult period. Let’s get started.
Facebook, although historically successful in removing 88.8% of hateful content, still struggle to identify COVID misinformation
In an attempt to combat the issue, the social media giant is partnering with up to 60 fact-checking organisations to do manual reviews. Now why would one of the world’s most tech-rich companies invest in such manual labour? I’ll point out the platform’s ability to influence and the event’s anomalous nature.
Facebook’s 2020 First Quarter Report states that it has 2.6 billion monthly active users (MAUs). That’s about a third of the world’s population! This is before we even mention MAUs in their family of apps, Instagram and WhatsApp. Even if we’d want to make an impossible assumption that half of their accounts are second accounts, they’d still have a user base of ⅙ of the world’s population. Pretty mad. Moreover, the report states that the increase of at-home bodies further catapults this metric. In this case, Facebook is its own establishment with Zuckerberg as its leader. Now do you see why misinformation becomes a crime in this virtual country?
Having established the power of Facebook and the issue of misinformation, let’s dissect COVID-19 misinformation specifically, and the general challenge of machine learning in disaster scenarios. The closest time-based occurrence we have to COVID-19 is probably the 2013–2014 Ebola outbreak or the 2002 SARS virus. Volume-wise, COVID-19 resembles the 18th Century Great Plagues and the 2009 Swine Flu. However, many of the technologies that we have now did not exist back then, including advanced transport, financial enablement, and data collection. Every epidemic or pandemic creates uniqueness that makes it especially difficult for algorithms to work accurately to fight them. Due to the lack of understanding, we as individuals and corporations have personal responsibilities to share sound information and avoid inappropriate responses.
Back to our point about providing actionable items — if you would like to understand more of how algorithms work for prediction, you may take Andrew Ng’s Machine Learning course on Coursera. For a challenge, compete in Facebook’s Hateful Memes Detection Challenge, or have a look at the UK Government’s COVID Simulation Code in C++, in collaboration with Imperial College London. Share with us some ways you think we as a community can move towards improved accuracy! In the case of COVID-19 related datasets, don’t be discouraged by low accuracy results, as the general machine learning field is still striving to tackle this compelling problem! Think better, not perfect.
Staying informed on insights from field experts about contact tracing, exposure notifications, privacy, and safely executing re-openings
We’re seeing a myriad of contact tracing and exposure notification applications in response to personal and government-level needs. Like the uncertainty that comes along with COVID-19 misinformation detection, there is no clear standard of what contact tracing or exposure notification tools should look like. Government initiated apps have been released alongside privately developed ones. The technologies behind each differ in privacy and optionality, amongst other factors.
To better understand COVID-19 mitigation tools, we’ll explore contact tracing and exposure notification apps, and the available backend options for each.
The difference between contract tracing and exposure notifications is that contact tracing allow for remote data storage. Essentially what this equates to is the ability for contact tracing apps to pass on information of a threatened individual to selected second parties, such as health officials. Conversely, exposure notifications only notifies potential COVID-19 threats to the owners of the device. Bluetooth, location detectors/GPS, data mining, decentralised privacy-preserving proximity tracing (DP-3T), and Apple/Google APIs act as optional backbones. The combination of systems limit or enable capabilities for other tools. It is up to developers and governments to decide on ethical means to mitigate this pandemic. You can read a more detailed explanation of the mentioned systems in MIT Technology Review’s article.
For more information on the implications of COVID-19 mitigation tools, sign up for TechCrunch’s COVID-19 Technology Task Force Forum. The event, co-hosted by multiple organisations including Harvard’s Berkman Klein Center and NYU’s Alliance for Public Interest Technology, will be held on the 17th of June. This forum will focus on recent contact tracing developments and available tools to help leaders reopen establishments safely. Speakers include MIT Media Lab’s Harper Reed, Microsoft Research’s Mary L. Gray, and former White House Deputy U.S CTO Andrew McLaughlin.
In addition to the MIT piece, TechCrunch has also published an article to help you get a headstart on privately developed COVID-19 mitigating tools. To sign up, head over to the official COVID-19 Technology Task Force event page. The discussions might help to wrangle with some of the previously mentioned datasets!
Seeking remote mental health help during this difficult season through apps like Calm, Headspace, and the COVID-specific Covid Coach
We wish you wellness and health during this season, but completely understand and empathise if the current circumstances have interfered with your emotional balance. We at Zero One also take measures to encourage each other through regular colleague check-ups, while maintaining social distancing. Each one of us have different hobbies to help cope and stay human during this period of time. However, we acknowledge the increased likelihood of one experiencing anxiety, stress, or even depression throughout this situation. Here are some apps that may help you get through social distancing in a healthy manner.
Mindfulness applications such as Calm and Headspace have existed before this COVID-19 period. These two apps in particular provide more general approaches to wellness through meditation and breathing exercises. More extensive programmes are offered in NeuroFlow and Unmind. These apps monitor your day-to-day routine and record activities such as sleep, mood, and stress levels. The information is then used to create personalised guidance tailored to your habits. Finally, COVID Coach is an app developed with specific features to combat pressing issues such as heightened levels of domestic and substance abuse during circumstances like these.
The aforementioned applications—as well as many others we haven’t mentioned—may not work for everyone, as therapy is still needed for exceptional and prolonged cases, but it’s possible that they’ll be your first step to self-help. You can download them freely or on subscriptions straight from your App Store or Google Play Store. You can also be encouraged by reading more on the impacts of wellness apps in this Carnegie Mellon article.
That concludes our first newsletter and we hope you’ll join us in our next one! Please let us know what you think and feel free to respond to the topics above. Stay safe, well, and healthy!