Trading Privacy for Value
In the start-up culture of the 21st century, we live by the motto “move fast and break things.” What if what gets broken is society*?
how can we build data products and services that use data ethically & responsibly?
how do companies take a data (science) project from lab to production successfully?
Systems that can explain their decisions.
how can we interconnect the web of data, its agents, and their decisions to enlarge the pie?
3. “… Advertising has us chasing cars and clothes,
working jobs we hate so we can buy shit we don't
need…” —Tyler Durden
http://kirkakahoshi.com/Finding-My-Inner-Tyler-Durden.html
7. • my friends and family, my shopping basket,
• my books, my photos, my music,
• my code, my meal, …
๏ I am all in!
Me and my, oh my!
https://www.octos.com.au/blog/4-marketing-lessons-learn-race-stops-nation/
8. Digital Me
• My cloud of information is used to construct a
Digital “Me”.
• Online advertisers advertise to “digital me”.
9. • What should we measure for representing users
“accurately”?
• We can measure (as in data collection) gender, age,
salary, occupation, location, device, browser, pleasure,
pain, …
“Measure what is measurable, and
make measurable what is not so.”
— Galileo Galilei
10. Measuring Happiness*
✤ Memory-based approach treats users retrospective
evaluations of past episodes and situations as valid
data.
✓ Moment-based approach derives the experienced
utility of an episode from real-time measures of
the pleasure and pain that a user experienced
during that episode.
* Daniel Kahneman. (2000). “Experienced Utility and Objective Happiness: A Moment-Based Approach” in Choices, Values and Frames, pp. 673-692, Cambridge University Press.
11. • Everything that can be measured and collected will
eventually be measured and collected …*huh?what!
> excuse me?
• Should we be worried?
Privacy
Photography Credit: Jack Wallen
12. Trading Privacy for Value
• “There is no clear right and plenty of well-known
wrongs. I think what is ultimately going to matter is the
notion of value for data, and for a lot of these large
systems, it will be value for privacy, if you will.”
• “Nobody wants to give up their privacy, even
incrementally, for vanity purposes, but if you’re giving
it up and realize that you’re giving up 10 pounds of
unwanted fat, or regaining three hours of sleep, then I
think people will just figure out how to cope with it and
move on.”
— Max Levchin, CEO of Affirm (PayPal Founder).
13. Problem of Trust
• We may have been abused without knowing it.
- Why was my insurance claim denied?
- Why was I not approved for that last loan?
- Were those decisions made by a system that was trained
on biased data, discriminating against me?
• In the start-up culture of the 21st century, we live by
the motto “move fast and break things.” What if
what gets broken is society*?
• Panel Question: is the problem of trust real, and if
so what are the consequences?
* https://www.oreilly.com/ideas/the-five-cs
14. Responsible Data Science
• We rarely talk about how to build a data product in
a responsible way:
- How can we treat others’ data as we would treat our
own?
- As a data scientist, have you actually thought about
how your own data might be used and abused?
• Panel Question: how can we build data products
and services that use data ethically & responsibly?
15. Team Sport
• Many companies struggle with getting value out of
their data projects, most of which stay as PoCs.
• Data Analysts, Data Engineers, Data Scientists, …
There is lack of unity, and failure to work together.
• Panel Question: how do companies take a data
(science) project from lab to production
successfully?
Spirit
16. Intelligent Systems
• Systems that can explain their decisions.
• Systems that can provide meaningful notions of
their uncertainty.
• Systems that pursue long-term goals, and actively
collect data in service of those goals.
https://www.awakeningfromalzheimers.com/peoples-brains-are-shrinking-at-an-alarming-rate/
17. We have been myopic!
• We designed (recommendation) systems that
interact with one user at a time, that make one
recommendation at a time.
• And these systems have grown up to be myopic; we
recommend
the same book to every reader,
the same movie to every kid on the block,
the same restaurant to every diner,
the same street to every driver,
…
Source: Thinkstock Images
18. • “We need to design systems that create markets
where data flows between producers and
consumers, and economic value can flow.”
— Michael I. Jordan of UC Berkeley.
• Instead of rich get richer, we can
have the pie get bigger.
• Panel Question: how can we
interconnect the web of data, its
agents, and their decisions to
enlarge the pie?
Data Economy
Photography Credit: Elise Bauer