Thoughts on Facebook Slingshot

Facebook’s answer to Snapshat, Slingshot, was released yesterday. The take on instant photo and video sharing is interesting with a strong direction towards increasing engagement in communication. For those who missed the principle Slingshot let’s you send photos (or videos) to contacts, to see the photo your contact should send one back. As with Snapchat the photos are deleted once they been seen.

As it happens, I’ve been supervising  Chloé Manceau‘s research thesis on multi-device communication and engagement. Chloé comes from the Information Architecture Masters in Lyon. She started by interviewing 8 people from teenagers to highly connected professionals. The main take away from these interviews was that engagement in communication was not binary (engaged or not). And thus that engagement does not mean instantaneous reply, but rather sustained exchanges. Based on these insights she designed a dual system to develop sustainable forms of engagement : A multi-device dashboard inviting to get back in touch with people whose conversations are “hanging”. And Missive, a messaging service with once a day delivery.

Chloé is defending today, thesis and videos are upcoming.

Observing and understanding usage at a distance

While preparing a post on Big Data for HCI, I started exploring tools for gathering usage data on the Web. There are a lot more tools than I imagined so I won’t list them all. Here is an quick attempt at categorizing them.

Descriptive tools

Descriptive tools give insights on users’ behaviors but do not enable to engage with them. They mostly focus on general stats and support some kind of user segmentation, but it’s mostly about broad quantitative insights.

Site analytics : Tools like Google Analytics and many others. Analytics are rather marketing oriented, but I talked to some startup developers who hooked Google Analytics to their mobile app to know which features were used and which were not.

Page centric analytics : More specialized tools like click trackers generate heatmaps and reports based on mouse events.

Controlled experiments

More widely known as A/B testing, web based controlled experiments have been used for the past 10 years but are increasingly popular. A/B testing enables to run controlled experiments to improve websites or mobile apps often focusing on a specific design point measure with a specific metric (conversion rate, money users spend, etc). An article explaining how to set up experiments (pdf), back in 2005 Jakob Nielsen wrote an article on the limits of the exercise. Google uses this (too?) widely which led some of its top designers to quit in response.

A/B testing tools include Google Web Optimizer, unbounce which specializes in landing pages, and many more. There is also a number of emerging tools for mobile apps such as Facebook‘s one.

Remote usability testing

Remote tests : Unlike A/B testing, remote usability tools enable to study at a distance how a (small) number of users are executing a specific task, often measured in terms of time and errors, sometimes with post-hoc defriefing. 

Remote walkthrough : tools like Ethnio support remote recruiting. You can then organize usability sessions by walking users through a task or interviewing them with a variety of tools.

Lightweight tests : a variety of tools support quick data gathering. For instance by asking for rapid feedback or asking what visitors remember after being exposed to a screenshot or mockup for 5 seconds,

Expert feedback :  designers ask for the feedback of colleagues/experts. Most often the feedback seems to be on screenshots so still pretty limited.


Nothing original here, just surveys served over the web…

Research on the topic

If you know of papers discussing the methodology of running studies (not only controlled experiments) over the distance, leave a comment or send me a ref via twitter @aurelient. Another post will cover research projects once I have enough material.

Online privacy and monadic principles

I’ve been trying to figure out how monadic principles (fr.wikipedia is actually richer) can help us think online privacy. I’m in the early stage of writing a paper on the topic, so this is really work in progress, input welcome.

The monads I’m talking about are not the functional programming ones, nor the mathematical objects, but their inspiration: the philosophical monad which could be defined both as the entity generating all other entities but also the all encompassing entity. Bear with me, for the technically inclined, know that Alan Kay mentions Leibniz’s monads as one of the inspiration for object oriented programming.

I’ll explore in upcoming posts how Tarde’s monads could be used practically to rethink the design of online socio-technical systems such as social-networks, but maybe also how to deal with online profiling/advertising tools.

Posts to come:

  • Introduction to Gabriel Tarde’s monadology via Bruno Latour
  • Heterarchy
  • Dividuel and online identities
  • The parts are greater than the whole
  • Humans and non-humans