The Obvious Invisible
Five Lessons for Predicting the Impossible
Around 10 years ago, when I was running up and down Sand Hill Road, none of the funders I had met with had predicted the launch of a billion dollar industry around voice-first products. The VC’s had all sorts of arguments about why what we were doing wouldn’t work, but not one could see the thing that seemed so obvious to me.
I was unskilled at communicating this future that was clearly obvious to me and that was heading towards us like a freight train with a loud horn that could be heard miles away.
Why did it seem so clear to me but not to them?
In the few years that preceded those pitches, I had buried myself in the books of the great predictors: Raymond Kurzweil, Alvin and Heidi Toffler, Kevin Kelly, Chris Anderson, among others. I also had started following the technology trends, looking at startups and trying to come up with ideas for my own ventures. This included a few failed attempts at ideas and a short gig limelighting as a CTO. It also helped that my day job involved visiting research labs and trying to sell them equipment that could help accelerate their work.
What came out this experience was the shell of a methodology that I couldn’t articulate at the time. It was a way of making it easy to see technologies and their influence. Since then, I’ve been thinking about this and how to make it easier for others to see what I sometimes see as the future.
The Loud Horn from Around the Bend
That freight train that was barreling down on our startup was the Amazon Echo. It would become the most successful hardware product launched by Amazon and outsold any competitor by a landslide. Google, Apple, Samsung, and others eventually tried to get into the smart speaker game and expand their footprint as smart assistants. However, at the time, this horn was drown out by all the other noise that was happening around the hardware space. There were WiFi plugs, thermostats, doorbells, and cameras.
Lesson #1 — There is always a lot of noise drowning out even the loudest signals.
You know this by looking at the headlines for technology today. It’s easy to see what tech news is dominating and where things are going in the short term. Sure, this is important but it’s mostly noise. There is almost nothing you can do with this news except be the first of millions to play with a new API or patch some really horrible vulnerability or see if your name or job was compromised by hackers.
To see if something really is a consistent loud horn, the signal needs to remain strong and grow in strength over time. A quick way to check this is to look at news from six months to a year ago. What has changed since then and what impacts could a new technology have on this news?
Sometimes the signal is that something that was just launched now and unlocks a technology that had little application six months to a year ago.
Doubling is Deceiving
Things that double in any way over anything less than a decade are very deceiving. Even after we learned the hard lesson of this during the pandemic, it’s still hard to fathom that if something doubles every week, if it starts at one, in a year it’s at 4.5 quadrillion. Think of the old Star Trek episode of the The Trouble With Tribbles to understand doubling. Cute, fluffy, doubling, deadly.
It’s the same with things that are affected by Moore’s Law. Everything technological doubles at some sub-decade interval until the limit is irrelevant. Whatever is the barrier today, whether it be processing speeds, memory, or connectivity it will likely not be an obstacle within a predictable period of time. The rates of advance in the limiting factors can be looked up within a few minutes of a Google search.
Lesson #2 — There are no limits.
To make the invisible become clear, you just need to ask limit removal questions like “what would happen if you had unlimited processing capacity?” or “how would terabit Internet speeds affect the experience?” Suddenly, using a 500 trillion parameter language model seems like something you could easily retrain in a second with new modeling data. When someone responds that “it sounds like science fiction” when you remove limits, that’s when you know you’ve just revealed something others can’t easily see.
Sometimes, there’s no trough
There’s a lot of hype around the Gartner hype cycle. It’s often used among different industries to show what’s the state of different technologies, what should be invested in and what should be put on the backburner. It’s a great tool for product managers to point to when deciding whether it makes sense to use a new technology or not. “Let’s invest in this” they say for things that have reached maturity or “it’s too early and this is just hype”. However, with Gartner there’s an overwhelming focus on technologies that are ascending or descending the hype curve and not on ones that have made it through the trough or are have reached some maturity.
After a casual reading of the hype cycle, one might conclude that all technologies follow this pattern. However, that’s not the case. The “Trough of Disillusionment” is often a graveyard and many early technologies die on the ascent of the hype cycle. 3D TV died, it’s not slowly ascending a Slope of Enlightenment. Likewise, many technologies follow a steep slope followed by a long plateau and the inflated expectations aren’t there because most users and business just don’t understand the capability of the technology. Crypto was really hyped because of currency speculation, not as much so because of its technological promise.
Lesson #3 — Hype and impact are not always related.
While hype can help spread the word and inspire others to build applications with a new technology, it’s not really an indication of whether the technology will have an enormous impact. What one can count on from analyses like the hype cycle are that some of the early technologies presented will have a game changing impact 5–10 years after their launch.
When looking at early technologies that come out, one can ask “what can this be combined with that will more than double its impact?” or “if this technology were fueled with a trillion dollar investment, how would its improvement be felt and what new applications could be built with it?” This can help weed out the bland applications from the exciting. You can place yourself in that 5–10 year future of a fully mature technology and see if it’s really exciting or not.
Back to 3D TV… “OK, so the video is seamless 3D and there are turbines that blow air when there’s wind in the movie or video game…” As Shania would say, “that don’t impress me much”.
Cheap and Ubiquitous
One of the driving factors around the Internet of Things revolution was the falling prices of WiFi and Systems on a Chip (SOCs). This made it easy for manufacturers to start to put Internet connectivity on their hardware that could feed sensor data to a service or allow for remote actuation.
Hardware makers are notoriously cheap. Every dollar added to a Bill of Material cost (BOM) increases the end consumer price by 3–6x. So, cheapness is king and whatever is added to the product must make it that much more compelling of a sale.
The same influence of BOM or its equivalent plays out over all new services and offerings of technologies. The trend of costs that make up a service or offering is for the most part predictable. Cloud computing prices drop, what seems exorbitant today can be predictably lower in cost in 2–3 years. All it takes to see a trend is two or three data points. So, if you start to see prices come down relative to inflation, then it’s worthwhile to look at all of the opportunities that might be unlocked.
Lesson #4 — Watch for falling prices.
When I was in school, one of my economics professors talked about how when oil reached $40 / barrel, it started to make economic sense to extract it from oil shale and that Canada had huge oil shale reserves. This lesson stuck with me in looking at technologies.
At what price does it make economic sense to start capturing sensor data in real time? What can be done with this data? What new insights can be gleaned that can be commercialized? Falling prices mean that new capabilities can be turned on that can be used to develop consumer or business desired features.
Expanding Branches
As the march of technologies continues and new tools get released, the tipping point starts to form when such technologies end up expanding on capability that is more than the sum of different parts.
Before we settled on a voice-based computer with our startup, we looked at creating a WiFi-enabled plug. First, it was just a sensor to measure power consumption and send this to a user on the Internet. Then, we added actuation to it and all sorts of capabilities opened up. When we were pivoting to the Ubi (the first voice computer on the market), we thought about adding a speaker to a plug. Then, we thought about what would happen if we added a microphone. It was obvious that when that combination of features was used, the number of possibilities expanded into the hundreds.
Lesson #5 — Look for the explosion of possibilities.
When the right mix of technologies come together, the potential applications will grow exponentially. That’s when you know you’ve hit on something that will be a game changer. You can also look ahead for when the different technologies that will make up that mix will mature enough to work together. You can then draw a circle around the technologies and come up with a prediction of how the future will look.
Being 100% accurate about the future is impossible. The unexpected will happen. If you apply a system of looking at when future technologies will mature and what can be done with them, you’ll be in a better position to see the “unforeseen” — long before others. The drawback of being so far ahead is that you’ll then have to develop some patience for others to come around to your advanced viewpoint. Such is the pain of the prophet.
