Email to my class about the Facebook et al. scandal
I teach Applied Machine Learning at the University of Maryland College Park. The class is 3 hours of lecture/lab every Wednesday. Today, I spent 30 minutes going through the scandals with Facebook, Cambridge Analytica, Microsoft banning NSFW content from most of their consumer facing cloud services, and China's Social Credit System, and discussing my concerns with each incident. Afterwards, I sent this email to my class and a colleague, who requested I publish it, and it's reproduced in it's original form below (hence the grammar and formatting).
Point of Facebook, Cambridge Analytica, China and Microsoft stories
When I went into the data and privacy stories today, part of it was just so you guys knew what happened, and part of it was to illustrate some broader points. I don't think I stated all of the key points as explicitly as I should've, so here they are:
-Data, particularly the nuanced profiles in social media websites Facebook and messaging histories, is incredibly weaponizable (largely via machine learning) and this has already been done
-To this end, giving someone access to massive amounts of personal data is a serious choice (I mean things like Facebook and Google, not cookies on a website or stupid things like that people complain about), and such companies need to be highly regulated and are not
-Machine learning isn't perfect, and the models intrinsically reflect our own biases (both in creation and data), and as such making important decisions (noteably including what information to show along with more obvious things) based on machine learning can't be taken lightly and is
-Using open source and very secure forms of communication is the only way to prevent the widespread weaponization of data.
-This is additionally important in that are without the proliferation of them you can't preserve any reasonable degree of freedom information or speech in the modern world (not the libertarian ideal of true free speech but many basic things any reasonable person would agree on) which is a fundamental human right
-I personally would not trust Facebooks services, which notably include Instagram and WhatsApp
-Signal is a relatively new open source messaging platform that's very similar to WhatsApp, is as secure as is reasonably possible (and enough to keep professional hackers happy), and as of major updates in the past 6 months can serve as a legitimate replacement to major messaging platforms. [It has no legitimate competition in this sphere, though as it's open source so it doesn't need it as much; the closest is Telegram which tries to achieve the same goals but has severe technical deficits, most notably regarding the quality of it's security].