AI and ML
Creativity is Remixing
This is a mind opener: "Image-to-Image Translation" and "Photo Style Transfer" give us a taste of how creativity can be automated.
They take in for example a daytime photo of a City A, nighttime photo of City B, and generate a nighttime photo of City A.
The input are static images and the results are good enough but not perfect, but imagine if you could replace images with audio, 3D, even designed products. Given you're able to create countless iterations, and train or review the results, even creative/critical thinkers can become obsolete (sorry!).
In the short term, feel free to use ML for Snapchat Filters or Smart Replies.
Product Design Weekly
Human Factors
How Uber Uses Psychological Tricks to Push Its Drivers’ Buttons - The New York Times
Last week I mentioned Dan Ariely helping companies with Behavioral Economics for good (saving money, etc).
This week, the NY Times published a 5000-word article about how reportedly Uber uses Behavioral Economics for profit.
I'll save you the 30min read and give you a summary of what I learned:
- Problem: Riders want a quick and cheap service, so the challenge for Uber is to create enough supply (Drivers) to ensure Riders get what they want, but not so much that Drivers are sitting around without a Rider. To complicate things, Drivers have no contract and no obligation as to when and where to be.
- Solution: According to the NYT, Uber employs 100s of Behavioral Economics professionals to come up with ways of controlling when and where Drivers work:
- Achievements: Recognizing achievements sustains engagement. Uber grants their drivers badges such as Above and Beyond, Excellent Service, Entertaining Drive;
- Loss Aversion: It's better to not lose $15 than to find $15. Uber tells Drivers "You're missing out on $15 by not working Fridays";
- Set Completion: The closer we are to completion, the more we desire to complete. Uber says “You’re $10 away from making $330 in net earnings. Are you sure you want to go offline?” [go offline] [KEEP DRIVING];
- Female Persona: Heterosexual men seem more likely to follow orders from women. Uber says "Hey, the concert’s about to let out. You should head over there. – Laura"
- Appropriate Challenges: We're more likely to join a challenge if we think we've got a chance to win. Uber has discovered that Drivers with more than 25 rides are less likely to abandon the service, so Uber sets that as a small achievable goal for new Drivers, encouraging them once they reach 50% of that goal (another form of Set Completion or Sunk Cost Bias).;
- Positive Feedback Loops: Visible progress is more Additionally, Positive Feedback Loops in the form of performance stats (trips/week, earnings, drive time, rating) keep the drivers going beyond
- Intrinsic Motivation: We're more likely to do something if we think it's our decision. Uber avoids explicitly telling their drivers what to do, as seen in the Set Completion example, Drivers can choose to [Go Offline (and miss the $330 goal)] or [Keep Driving].
- Image Recognition: According to the NY Times, Uber has 2 visually similar buttons: Surge Pricing and Higher Chance of Surge Pricing. Drivers claim the latter appears often but seems designed to be hard to dismiss and doesn't live through its promise.
The real kicker is that Uber responded to the NY Times article claiming they do use Behavioral Economics in this fashion but for the benefit of both Drivers and Riders.
Given neither Uber or Lyft make a profit just yet, and both seem to be playing a variation of the Embrace Extend and Extinguish game, the question here is: Is Trust or Behavioral Science better in the long-run?
nytimes.com
Chatbots
ChatBots
Summary
- Facebook Messenger
- Beats Apple in the race for a proactive Virtual Assistant which inserts itself into conversations, it's called M and it's broadly available to anyone in the U.S., including that guy soon to be impeached. 🇺🇸
- Making it easier to figure out what exactly a bot can do for you by giving the option to take the Chat out of the Chatbot and only allow input through a hierarchical menu;
- Bringing chatbots to group conversations, perhaps to beef up against Slack, diversity and increase awareness;
- Google Assistant rolling out to more Android phones Still unsure how that benefits us Product Designers. It's capable of acting on in-app events/media/more but sounds like a something you'd use on a very specialized application.
Design
Subform | A modern tool for digital product designers.
"Spend more of your time designing—instead of juggling multiple artboards and tediously propagating changes everywhere."
When we create a product, every screen is unique as a whole but composed of many reused parts, colors, copy, and more. There is also a before (research), during (testing) and after (release). Yet, there is currently no design tool (or development for that matter) that has been designed with all of this in mind, so it's good to see more work being done on some of these issues.
subformapp.com
MLB.com At Bat and NHL Apps Take Advantage of Custom Icon Option in iOS 10.3 - Mac Rumors
If you were wondering about valid use cases for this recently introduced feature.
macrumors.com
Education
How cloud computing has changed homework time—for parents | Ars Technica
- Cloud computing is not an issue for low-income families, it simply means they use smartphones instead of desktops (some of those who can choose actually prefer smartphones)
- Teachers don't see cloud computing as key for their performance: "If it all went away, I'd still be able to do my work"
- Parents feel uncomfortable with the reduced visibility and control over their kids homework
- Good Docs is a professional tool that solves document editing and sharing, but it's not geared towards schools who still struggle to maintain a child's portfolio (article doesn't specify what type of portfolio: document list or historical performance?)
History
Process
Our approach to innovation is dead wrong | Diana Kander | TEDxKC
"Everyone has a plan 'till they get punched in the mouth." – Mike Tyson
Diana talks about Waterfall vs Lean resorting to a Mike Tyson quote and the Marshmallow Challenge.
The Marshmallow Challenge … constructing the highest possible free-standing structure with a marshmallow on top … within 18-minutes using only 20 sticks of spaghetti, one yard of tape, and one yard of string.
TL;DR: Education makes you risk-averse and over confident, only kindergarten kids succeed because they don't spend most of the available time planning and thinking they only need one shot at it (Waterfall), instead they try out something safe (MVP) and build upon it (Test/Iterate).
youtube.com
Reality
Watch this guy catch a real ball in VR - The Verge
Highlight of the week: imagine you could see where an object will fall as it moves in the air.
theverge.com
Research
Filling Up Your Tank, Or How To Justify User Research Sample Size And Data – Smashing Magazine
Great article about coming up with a formula for participant count in Qualitative User Research.
Summary:
- Scope: Complexity-based multiplier – Improve (1), Create (2), Generalize (3)
- Population: Minimum 3 participants per user type
- Expertise: Expertise-based divider – Junior (1), 5 years (1.10), 10 years (1.20)
- Resources: Lower or increase the participant count based on available time and money, be clear about it. e.g. $30k and 2 weeks available, lab+hiring for 15 participants would take $45k ($3k/head) and 3 weeks (2days/head) respectively, resources should have a value of -5 to lower the required 15 to 10.
`` ( ( Scope x Population ) / Expertise ) + Resources = Participants
This will certainly generate some discussion, the article has more details around data saturation, worth a read.
smashingmagazine.com
Security & Privacy
Here’s all the new stuff in Apple’s latest security document | TechCrunch
Not for the common mortal, but keep in mind such a resource exists, it's a RTFM for security-conscious people. Updated to explain the security of latest Touch ID, HomeKit, ReplayKit (screen recording), Siri, and Apple Pay features, in addition to sensitive technologies across all Apple's products.
techcrunch.com
Strategy
Facebook pivots into Stories
Stories are interesting because:
- More Content: Social networks work both ways, users share content and signals are shared back to them. These signals are the reward for sharing content and satisfy our craving for status and social validation i.e. "Holiday photo with 138 likes and 23 comments". The need to spend time curating content before it's posted is documented starts early and is the main detractor for social network engagement and growth. Instagram and more recently Snapchat reduced that friction through filters that add value to otherwise low-value content. Ephemeral posts enable people to post lower quality content without fear of long-term repercussions.
- More Data: More engagement means more data, the neatest trick Snapchat pulled was to convince its users to share their location with the company through Geo Filters. By tracking where users go after being exposed (or not) to geo-tagged posts, Snapchat is able to measure the conversion rates of those posts in what they call "Snap-to-Store", bringing it closer to Foursquare in terms of Place Analytics.
Voice UI
2017 Voice Report | VoiceLabs
Summary (I wouldn't take this source too seriously though, good enough to get a pulse)
- Voice UI support in Services (e.g. Voicelabs.co), Applications (e.g. Philips Hue), AIs (e.g. Siri), to Hardware (e.g. Amazon Echo);
- Google Home & Amazon Echo shipments grew significantly in 2016
- Roughly 7000 Alexa Skills (Amazon says 10000), mostly in News, Games/Trivia, Education & Reference, Lifestyle and Weather
- Most people use their Home/Echo to listen to Music, Control Lights, Games/Trivia, News, Set Timers
- 3% retention rate for Home/Echo users (novelty)
- Little to no monetization
The Matrix Voice is an Alexa for your Raspberry Pi
Great for prototyping context-aware VoiceUIs as it includes 8 different sensors (humidity, ultraviolet, temperature, motion, more), and 4 wireless communication channels (NFC, ZigBee, Z-Wave, IR). The underlying Raspberry Pi means it's very extensible and can be made mobile.
techcrunch.com
In this issue:
- Turmoil in advertisement land: advertisers want more control, some pulled out of Google in retaliation (it's always hard to have a product and please all customers);
- A list of Behavioral Economics "tricks" employed by Uber to manage their drivers, pulled from a 5000-word NY Times article: we do the hard work so you don't have to;
- Small insight into how cloud computing is changing Education
- An attempt at figuring out sample size for Qualitative User Research
- 17 years of Mobile Handset Manufacturers in one graph
- What are Stories, why is Facebook so keen on them, and how Snapchat is doing place analytics
- Some updates on ML, Virtual Reality, Chatbots, Voice UI, Tools, Innovation Process, Security, and use case for iOS 10.3 Alternative App Icon
- Cats and Bells 🐈🛎
Luis AbreuTurmoil in advertisement land.
Never expected to get these many ad-related links in a single week, scan these out: