Thursday, July 29, 2010

Future Shock Applied: internet of stuff

The internet of stuff is coming soon.  This is a build-it-and-they-will-come project.  I've never heard anyone give me a convincing argument for why the internet of stuff is coming.  But lots of people are working on it, and clearly it is coming.  In this article I will explain both what the internet of stuff is and why it is coming.  I will also explain what value people will derive from it, and how that value will cause subtle but major market shifts.  If you fail to understand this then you will find yourself on the wrong side of of those shifts, and you will be replaced.

What is the internet of stuff?
The internet of stuff (IoS) is little computers embedded in every device that could possibly hold one.  Some of these computers will be little more than tiny sensors that report "I'm over here."  Some of them will be robust, and allow things like direct internet access--like the newest LG refrigerator.

All of these computers will connect to the internet.  Some of them will actively broadcast status.  Some of them will passively wait for you to query them.  And they will talk to each other constantly.

The researchers at Maya are working on this for Darpa.  They have a good introductory video, here.  They call it 'pervasive computing', but it's the same thing.

Some examples are probably in order.
* Nike has shoes available that have sensors that connect via Bluetooth to your iPhone.  You can run a pedometer application on your iPhone and the sensors on your shoes will keep track of your steps and speed.
* Your electric meter has a small computer that sends usage statistics to the electric company.  It sends this data daily, right now.
* In the future the electric company will be able to reach out to a group of electric meters and order them to reduce their consumption.  Those electric meters will then contact devices in the homes (or businesses) and tell them to turn off.
* GM's OnStar already has crash sensors connected to its phone.  GM doesn't advertise it, but OnStar is probably also connected to the on-board maintenance computers and sensors.  GM should be tracking things like low oil pressure warnings, especially in leased vehicles.

Companies won't use the brand name "internet of stuff".  Generally speaking, they won't call attention to the fact that the device is connected.  They will focus on marketing the features that are built upon the connected computer.  The computers, sensors, and connectivity give everyday products new features that should make them better than the old versions.  But quantifying the differences will be difficult.  So they will use terms like "smart", "intelligent", or even "learning".  So you will hear about "smart toasters" and "intelligent wallets."

Why is IoS coming?
There are currently three different pressures causing this to happen.  Once the ball gets rolling a fourth pressure will eclipse the first three.

There is simple market competition.  Manufacturers of all sorts of devices are actively searching to add features that customers might appreciate.  The key word is 'searching'.  They experiment and adapt.  This is how the free market works to drive product evolution.

There is pressure to make devices more energy efficient.  This is s soft, social pressure among the manufacturers; but it is real nonetheless.  Sensors that gather usage statistics and patterns are an important step towards making devices that use less energy.  And those sensors are a first step towards making devices that can turn themselves off when they are not needed, which is the single greatest energy-saving feature possible.

There is also creative pressure within the designers.  The people who design devices want to do new and interesting work.  This type of connectivity is new and interesting.  This is just a fad, really.  But we are on the front edge of a fad that will last several more years.

It is important to note that consumer demand is not one of those pressures.  This is a build-it-and-they-will-come trend.  It is always the more expensive brands that add new features first.  The build-it-first nature of this trend means that it will be even more selective brands who invest in this trend first.  The good news for these cutting-edge companies is that they will gain a large first-mover advantage from the data they collect--as long as they have the vision to react to the data they collect.

For instance, the first company to build a smart desk chair will have the first collection of information about sitting.  At about the time that their competitors start developing their first generation smart desk chairs, the first-mover company will be designing their second-generation smart desk chair.  The first-mover company has all of the data that they collected to guide their design, and none of their competitors have that data.  If they have intelligently decided what data to collect, then that investment will pay huge dividends in competitive advantage.

And in this way the collection of data will become an arms race among competitive companies.  This amounts to a fourth pressure, but this pressure hasn't arrived yet.  However, when this pressure arrives it will greatly overshadow all of the other pressures.  Once companies identify the data arms race there will be no going back.

How long until IoS arrives?
The bleeding edge of IoS has already arrived, as evidenced by my examples above.  There are several obstacles to the full adoption of the IoS.

The sensors and small computers are still expensive, in component terms.  For instance, you can't add $60 worth of electronic components to a $30 toaster without raising the price to the consumer.  And consumers won't pay that much premium.  In most product markets no one yet knows how much premium consumers will pay for intelligent features.  But the free market is great at reducing costs as volume increases.  So this problem will self-solve naturally as the other hurdles are overcome.

IPv4 is about to reach the end of its capacity.  We are going to run out of IP addresses (the street addresses that computers use to find each other on the network.)  That's going to happen in the next year or so, even if we don't have a huge spike of internet devices added.  But when we run out of IP addresses there will be a mini-Y2K panic as internet expansion hits a wall.  There is a solution called IPv6.  It's mostly just a software change.  But everyone is postponing it until they absolutely have to do it, because it is expensive and could introduce some errors.  Our hand will be forced soon, and there will be a period of interruption during the real transition to IPv6.  This interruption will reverberate through the manufacturing sector because there will be a period during which consumers will not pay a single penny for added intelligence--because the new devices cannot be connected.  The interruption period itself will probably last 6-9 months.  The echoes in the manufacturing sector will last another up to 3 months.

But when that interruption period is over there will be a major rush to market for new intelligent devices.  Consumers demand will be pent up, and the premium they are willing to pay will go up significantly.

There is a lot of infrastructure that must be built, beyond just the transition to IPv6.  Companies have to decide what data to collect, how often to collect it, and then design databases to hold this data.  The database problem here is not trivial, because the volume of data will be tremendous.  As these systems hit mainstream the data loads will dwarf all previous data collections.  Existing database technologies are not ready for this.  There are a few cutting-edge database technologies that might can handle it, like Algebraix (disclosure: a friend works there.)  The first collection systems will rely on old database technologies, and will only last for a year or three before they are overwhelmed and replaced.  So the database problem will take a few years to fully sort out, and then there will be a rush of companies jump on the bandwagon.

The big non-issue here is privacy.  The Washington crowd will make speeches and hold hearings about consumer privacy.  Some luddites will buy an intelligent device and then feign ignorance that their private data is being collected, and sue.  The first luddite will win a huge settlement.  The second luddite will get laughed out of court.  There will be a lot of noise, and the lawyers will smell blood.  But the general public just doesn't care.

What value will people derive from IoS?
Of course there will be distinct valuable features for each product.  For example:
* A smart toaster will use bread-type sensors, bread-width sensors, and a feedback panel to learn how to not ever burn your toast again.
* A smart water-heater will use iron content sensors to understand the amount of rust happening internally, and will text you before it develops a leak.
* A smart washing machine will use fabric and stain sensors to decide on water temperature and soap usage; and will alert you when treating one stain will cause another stain to set.
These are valuable in and of themselves.  Every category of product will have new features like this that add genuine value.

But beyond that there is a special meta-value that will come along for the ride.  Consumers will eventually get hard numbers on reliability and maintenance costs.

How long will each of these products last?

How much will I spend maintaining each of these products?

How much time will I spend maintaining each of these products?

Today consumers can only guess about this information.  In my lifetime this information will become ubiquitous.  By the time my grandchildren are adults, they will have this type of information at their fingertips for every product decision.

From this side of the inflection point we have trouble understanding the value of this data.  I believe that the value of this data for consumers will greatly outweigh the accumulated value of all of the unburnt toast, un-flooded floors, and un-set stains.

Truly smart products will add their maintenance schedule to your personal calendar and budget the moment you buy them.  And they will remove themselves from your calendar and budget when you give them away.

How will this value change the marketplace?
The modern era has been identified with a long trend away from permanent products and towards temporary products.  Toffler called the future a throw-away society.  He lamented the expansion of the landfills and the emotional disconnect caused by throwing everything away.  Disposable paper wedding dresses were his favorite marker of a society that no longer cared about the value of their possessions.  Reliability and maintenance information will end this trend.

Consumers will finally be able to make informed decisions.  And this information will enable them to make better decisions.  Purchases will naturally fall into four big categories for consumers:
* Disposable products (items that are obviously consumed with use), like toilet paper and dish soap.
* Discretionary products (items I use until I replace with something better), like cell phones, automobiles, cheap furniture, and computers.
* Wear-out products (items that I expect to use until they wear out), like shoes, water heaters, and shingles.
* Semi-Permanent products (items that expect to keep for the rest of my life), like jewelry and good furniture.

The categories are arranged in order of increased expected useful life.  Reliability and maintenance information will have more impact on the longer-lived products.  This impact will be felt in three aspects of the various products:
* Market lifespan
* Price
* Warranty

Disposable products -- No impact.

Discretionary products -- Products with very poor reliability will disappear from the market quickly.  Maintenance costs do not really apply to this category.

There will still be crap products introduced into the market, but the early failures will kill products quickly due to plummeting real-time reliability scores.  This will force quality manufacturers to spend more time on product testing before release, because early manufacturing errors will have enough impact to kill otherwise great products.

Warranties on these products will be reasonably accurate signals of the manufacturers' expectations of reliability.  Manufacturers who build highly reliable products will ramp up their warranties (and prices) to suit.  Manufacturers who build crap products (either intentionally or accidentally) and provide competitive warranties will be punished with high warranty costs.  Therefore, new products will usually be introduced with lesser warranties (and lesser prices), until they build a reputation for quality.

Wear-out products -- Products with very poor reliability or high maintenance costs will disappear from the market quickly.  Products with slightly below-average reliability or high maintenance will sell at substantial discounts, and will only be situational purchases.

For instance, shoes that have poor reliability ratings (or high maintenance costs) will only be purchased for special occasions, when the buyer genuinely does not expect to use the shoes for very long.  They will effectively be disposable purchases, and will have to be priced accordingly.  These products will often be items that were designed to be higher-priced, but that had unforeseen problems and ended up in the discount bin.  They will mostly be discontinued sell-through products.

Warranties for wear-out products will follow the same patterns as discretionary products.  Lesser warranties on new products will be coupled with lower prices, reflecting the risk of not having established reliability and maintenance histories.

Semi-Permanent products -- The whole range of products will be available, with reliability and maintenance information figuring significantly into the prices.

For instance, cheaply-made furniture will be priced significantly lower than well-made furniture.  This is actually only a small change from the current market for these types of products.  Currently reliability and maintenance costs can only be guessed at, and prices are used by sellers as signals about these hidden qualities.  This information imbalance gives sellers an advantage.  Sellers will lose that advantage, and will have to alter their prices based upon the real-time reliability and maintenance information.

Warranties for semi-permanent products will follow the same pattern, but the differences will be even greater.  New products will have inconsequential warranties, and will have to be priced accordingly.  Once a semi-permanent product has established a history of reliability and low maintenance costs its price and warranty will rise.  This will put pressure on manufacturers to introduce new products less often, and instead to rely on sales of established products.

Consumers will be dramatically better off because of the improved information.  In economic terms, purchases will be made with less asymmetry of information.  That is the real value that consumers will receive from the internet of stuff.

No comments:

Post a Comment