Jeeves Must Go

There is a classic “Smart Home” scenario that you’ve all likely heard: my calendar talks to my alarm clock, allowing me to sleep in an extra 30 minutes. When I finally do get up, the alarm clock starts the coffee maker so as I walk into the kitchen, I’m greeted by fresh brewed coffee.

I hate this scenario with the heat of a thousand suns.

Reason 1: Humans aren’t algorithmic
22 year old single engineers may have a very predictable life but any parent will laugh at this scenario: their kids just aren’t going to sleep in. Even if you are single, there are a gazillion reasons to get up that have nothing to do with your calendar. And even if you had a regular schedule, there are always those random meetings, put in by others, which a human can easily ignore, yet would trigger an early unwanted alarm. It may not happen often but even if it triggers just once a month, it would still drive you crazy. Finally, let’s not forget those mornings when you’re in a rush and don’t have time for coffee. But don’t worry, the alarm clock will still doggedly make it for you…

It’s easy to get excited by the potential of automated devices but just because we can do something doesn’t mean we should. As a UX designer, two things has been drilled into me throughout my career:

  • Most humans are messy balls of chaos (this is a good thing)
  • Even if you’re wrong just 10% of the time, your customers will hate you

Good UX design isn’t just about making things work, it’s also about making things not fail. Unfortunately, most home automation vendors are just thrilled that things sort of work.

Reason 2: Don’t apply digital rules to analog tasks
Yes, the alarm clock can ‘turn on’ the coffee maker, that’s technically not that hard. But who replaces the filter, loads up the coffee, refills the water, and most importantly, puts a pot under the spout? The majority of the work on the coffee machine isn’t the digital ‘turn on’ command but the very analog ‘refill’ command. It’s far too easy for the coffee (or the pot!) to not materialize through human error and I can promise you, the human won’t be blaming themselves when they wake up to a coffee puddle on the floor.

Reason 3: The value just isn’t there
With all this technology at our disposal, is our big dream actually saving a few minutes brewing coffee? This seems like a profoundly modest goal. We can shoot much higher.

Jeeves must go
We are so smitten with the idea of an old school butler that we want a digital one. This service envy is the core driving force of most home automation scenarios. Having the low level tech to make things turn on remotely is a fine start, but we seem to forgetting a teeny tiny little point: we haven’t invented practical AI yet. I know, I know, it is coming: but it’s not here now. My point is that people seem to think simplistic if/then rules are somehow going to be enough.

Our preoccupation with creating our own personal “Jeeves” must go. It’s blinding us from thinking bigger and more interesting thoughts. Most home automation systems are hopelessly rigid and while they will give a great demo these systems are unfortunately doomed to fail too often. You can only say “Sorry honey, that shouldn’t have happened” so many times before the entire system is banished from the home.

I’m actually very bullish on the ultimate potential of the smart home, I’m just saying we need to look beyond Jeeves. Here are a few are simple and valuable things we could be doing right now that are not servant motivated.

Value Exploration 1: Simple status
Devices that come with their own connectivity, such as GSM modems, allow devices to upload status information to servers with zero configuration and no interaction with the customer at all. GE is already doing this with their ‘industrial internet of things’ and the valuable insights gained from this approach will come into the home as the cost inevitably falls. There are a range of solutions possible with this approach:

  • Internal Company Insights
    This would allow a company to gather data on their products and learn how they are used and perform in various environments. I realize you’re rolling your eyes at me but honestly, this is a big deal and companies would love to have this information if they could get it cheaply.
  • Consumer relationship
    The next step is to share this status information with the consumer, warning them of pending failures and basic maintenance (such as replacing filters) Of course, there is a risk that companies will abuse this power, trying to push replacements too often but that’s basic UX design and companies that get that right will gain far more than filter replacements, they’ll gain trust and brand loyalty.
  • Automated repair
    If the consumer agrees, the manufacturer can warn a repair company nearby that your water heater is starting to fail and can proactively call to schedule a repair before a catastrophe happens. This has been previously discussed by Jonathan Stark as “Setting the Whack”.
  • Quantified House
    Assuming we can get this status data for all my devices into a single centralized location (I know that’s a big ask), we could start having a quantified house, measuring electricity, heating, and appliance use. As Tesla’s Powerwall gains wider use, consumers will be VERY interested in where their power is going and a house wide audit will allow easy analysis and savings on electrical power. However, there is likely much more we can learn about how people use the home, what affects air quality, and ways to improve heating. These are all a bit analytical and not nearly as exciting as ‘attentive devices’ but there are some deeply valuable things hiding in this data.

Value Exploration 2: Lighting
I’ve explored lighting before in a previous post but it’s worth pointing out the range of possibilities that lighting enables and it’s far more interesting than the Jeeves inspired ‘turning on the lights when you enter a room” scenario which is not only a bit boring but also problematic (it’s far too easy to wake people up with automatic lights).

  • Just lowering costs
    A home lighting system using mesh networking will be much more flexible and less expensive to build. I realize this feels very dull but money is a big motivator and having a system that allows you to have much more flexible yet lower cost installation can be a strong product driver.
  • Simple targeted solutions
    Instead of monolithic systems that control the entire house, simple stand alone solutions are useful. For example, Ikea sells a battery powered closet light for just a few dollars. The problem is that you must manually turn it on/off each time you open the closet door. Just by adding a simple wireless door sensor, you could easily set up one (or even 20!) of these things to turn on/off when the door opens/closes. This is a simple, clean value that doesn’t over reach but solves a clear need. If this example doesn’t seem that valuable to you, just imagine if you had to manually turn the light in the refrigerator on and off every time you opened the door…
  • Impossible lighting
    It will be possible to have dozens of lights controlled with a single wireless switch. The ability to just plug in a new lamp and have it auto register to ‘the room’ so the switch can turn it on and off is one of those tiny things that can change the way we use lighting.
  • Safety Lighting vs Task Lighting
    I’m looking forward to a home with unobtrusive baseboard lighting that turns on and off entirely based on simple motion sensors. That is indeed a simple if/then rule but as it’s such a harmless consequence, that can be automated. The big bright ceiling lights will still be controlled by wall switches as that still gives me predictable control. However, these switches will be wireless ones which can be placed anywhere and more liberally around the house. In addition, I can still use my phone to remotely control everything if I need to but that would be a more exceptional use.

Valuable isn’t sexy

These examples are very valuable but, honestly, they just don’t demo as well as a magical digital servant. While a system that magically anticipates my every need is a lovely idea that will always aspire, it’s a bit TOO exciting actually, distracting us from the pragmatic things right in front of us. We’ll only discover more creative solutions if we drop our obsession with servant style automation. There are lots of simple but powerful concepts we can explore, build, and use today.

It’s time to let Jeeves go.



Special thanks to @jonathanstark for reviewing early drafts