Does Tech Have an Early Adopter Problem?

Or: “The Hype Must Cycle”

The exact moment is different for everyone. It might have been someone on a commuter train, wearing a virtual reality headset and plucking at invisible app icons. Maybe seeing someone simulate being underwater near a tropical beach while editing stock video footage in Final Cut Pro, with replies suggesting that people would love to be able to do this “for real.”

To me it was this: seeing someone drive around a suburban neighborhood in a Tesla, shooting a video explaining how to use an Apple Vision Pro in a moving vehicle. Not while sitting in the passenger seat, mind you, and not even while failing to pay adequate attention to Tesla’s sometimes dodgy autopilot, but actually driving the car. 

These are not just the sorts of experiences that make a person cringe or shrug with irritation at the silliness of it all. These are the sorts of experiences that create animus. They are the kernels of a grudge. A growing and overwhelming sense of alienation that attaches not just to the type of person who would engage in this kind of behavior, but to the thing itself. 

“Surely,” you ask yourself: “surely the product is as much to blame for this heinous abuse and misuse as the person doing the abusing!” That kernel of animus, given form and volume, begins to grow, and is hard or impossible to shake. It is now a part of your mental framing of the product. Not quite socio-economic grievance and not quite not; not quite Ludditism and no quite not, this framing persists and evolves with the product. 

Here in, I suspect, lies a problem.

To understand this problem, which I’m going to call the “Early Adopter Problem,” and which I don’t wish you to confuse with the Early Adopter Syndrome, we need to talk a bit about hype, economics, and the economics of hype. 

We’ll start with something just about everyone has heard about hype, but which many of us (by no means all!), have failed to seriously examine. 

The Hype Cycle

It’s almost certain that if you’re reading this post, you’ve seen something like the following image: a rendering of the famed Gartner Hype Cycle, a paradigm visualization popularized by Gartner, an IT advisory and research firm: 

The whole conceit of the Gartner Hype Cycle is that technologies reach maturity and create real economic benefits through mainstream adoption and ensuing productivity gains after following an S curve of “hype,” or public -and investor- interest and excitement about the technology. Note: this doesn’t apply to an individual product, but rather to a technology as a whole. It’s supposed to go something like this: 

Technology Trigger

The technology is supposed to be triggered when an early proof-of-concept, or even a fictional portrayal of a future technology generates excitement among early adopters and the media. The trigger is rarely a product in itself. It could be a white paper, a book, a sci-fi film, or these days, even a YouTube video. Anything that captures the imagination of enough people that the idea is taken seriously by a larger group of innovators. 

Peak of Inflated Expectations

The added attention leads to innovators and dreamers taking an early stab at a proof of concept. One of these early proofs may trigger a huge mainstream media and world-of-mouth response. For the first time, people outside the early adopter group become aware of the technology. Pundits and popularizers begin to make promises of disruption and game-changing benefits to the new technology. Often these early promises have little to do with reality. 

Trough of Disillusionment

Then the downturn begins. Early products in the category fail because they are too clunky, too expensive, don’t do enough to justify their cost, or just plain don’t work. In some cases, the reputations of whole companies become tarnished by the failure, and early innovators pay the price for these mistakes. These failures generate negative media buzz, and many early innovators fail at this point, while investors are attracted to the companies and products most likely to survive the cycle. 

Slope of Enlightenment

Now, with only a tiny foothold in its potential audience, but a market that is now aware of the technology as a concept, it begins a period of gestation. Second and third product generations serve a small audience of believers, and this group works to define the technology’s real benefits and the core use cases that it will serve in the future. Often at this point, innovators identify a more realistic market opportunity for the technology. 

Plateau of Productivity 

Finally, our technology has reached a point at which its use cases are proven, and its economic value has been recognized by a mainstream audience. It reaches its full potential with a target market, just in time for a new disruptive technology to be introduced, and for the cycle to repeat itself. 

Does This Actually Make Sense? 

If we take this paradigm as a guide to understanding events, we should be thrilled not when a technology first succeeds at gaining widespread media attention, but instead at the failure of those efforts aimed at early adopters, because their demise signals a maturation of the innovation space. 

We could look, therefore, on the recent catastrophic failure of Tesla’s Cybertruck, not as a signal that people didn’t want a rust-prone, flat-paneled fingerprint magnet, low range, small capacity electric pickup truck with no interior gauges, knobs, or buttons. Rather, we should see this failure and the attending media frenzy over every detail of the disaster as a signal that something great is sure to come in the long tail of “enlightenment” and “productivity.”

Perhaps one of the reasons the Gartner cycle is so popular in the tech industry is this promise: that all failures are actually the necessary seeds of success. That, and because the S-curve can be applied on any imaginable scale, with the visibility axis being applied just to one company or product, and the time scale being measured in anything from months to decades.

In this framing, the bigger the PR and business catastrophe, the more impactful the eventual plateau will be when it comes. After all, the long slope of productivity necessarily follows in proportion to the initial hype. The bigger the hype, the higher that plateau.  

Essentially none of the points on this scale actually occurred in any meaningful sense. The marketing industry didn't so much reject blockchain technology as remain categorically untouched by it. 

The Hype Cycle could well be argued to be more relevant as a description of the cycle of technology and business apologetics, than of innovation or hype. 

Right now, executives like Sam Altman of OpenAI are trying to thread the needle between huge promises about the imminent new capabilities of Ai, and a broad backlash against their business practices and mission. They surely would like for the broader public and business community to buy into the notion that the level of hype for GenAi is only time-honored proof that the eventual productivity gains will be enormous. 

Perhaps they will be. But will that eventual payoff really be a vindication of the Hype Cycle paradigm? 

The Cycle: Disconformity and Discontinuity

As you may have guessed, we’re not entirely convinced that this paradigm is either accurate or useful for understanding the real world, or the adoption of technologies in it. As the Economist recently noted, the data simply do not support it. Their quantitative analysis of a broad range of technologies showed that the cycle is at best a statistical rarity, applying in broad terms to only about a fifth of technologies included in their analysis.

A thoughtful critic could justifiably suggest that a 20% success rate is a poor showing. 

The number of actual cases in which “hype” is an essential ingredient in the development of a technology with great productive potential is remarkably low, all things considered. Technologies that enjoy disproportionate publicity fail all the time, and don’t always come back. Look about you and see if you can spot any of the following: a 3D television, a wifi-connected juice crushing machine, a Theranos blood tester, a Segway, the Metaverse, or a Hyperloop station.

Often these technologies failed, despite the hype, because they were never really based on a technology change or advancement that offered real productivity gains - gains that would never happen without them. On the same token, the cycle is often not applied to technologies that grow from minor to widespread adoption, at any pace, without a period of overpromise and under-delivery.

 There was no real S curve in the development of market expectations for cellular telephone technology, domestic electricity, internal combustion engines, airplane travel, satellite communications, or even the internet. No period of huge uptake in interest from innovators followed by a period of dormancy. Of course many small firms did start and fail in those industries, but this seems much more the product of a tendency toward monopolistic control under capitalism. Small firms lack market penetration and capital, so they either get bought up or sold. The number of firms operating is not a reliable indicator of public interest or expectations. 

Belying the cycle, many transformative technologies follow a rather flat curve of advancement over a given period, regardless of popular sentiment. 

[Credit: Scott Brinker: chiefmartec blog] 

After all, the factors behind technological advancement are rarely limited to media visibility. In fact they rarely are. Capital expenditures, fundamental research, and investments in commercialization rarely run apace with media narratives, as they are tied to anything from government policy to the broader economic cycle, to geopolitical factors.

 It’s only really in the SaaS businesses that were spawned in the great expansion of the public internet from the 90s to the 00’s that this cycle could even be considered important, because for the first time, the “hype” and the actual success of new firms was actually quite closely correlated, thanks to the newfound influence of search engines, and later social media. But underlying all this, many fundamental advances simply happen at a rate that follows wave after wave of investments of private and public money and research. 

Anything from WiFi to mRNA vaccines has failed to adhere to this hype paradigm. Instead, such technologies seem to come almost instantaneously into public attention just as they become available (note the emphasis on “seem”). Unseen are the large-scale efforts of regulators, industry consortia, and publicly funded research groups that create the basis for their widespread adoption. By the time the public has been aware of these innovations, they were already many years and many dollars in the making. 

Broad-based efforts like these preclude the likelihood of any boom and bust cycle. Institutional support and the capital investments behind them make their failure either unlikely or at least a distant prospect. The adoption of tech like fission/fusion power, radio and television, satellite internet, or cultivated meat have been governed far more by long-term investments and regulatory actions than by what people thought of their usefulness at any particular time. 

It’s possible for hype to modulate that cycle, to a degree. Surely the hype for cultivated meat products is going to be meaningful in getting people to actually buy it… but how meaningful? Technologies like cultivated meat have faced legal and industrial challenges for many years before they ever reach consumers at scale, at which point, as many in the industry predict, a simple consumer choice will drive the economic outcome: if cultivated meat is perceived as just as good, but is 20% cheaper, then consumers will buy it. More importantly, if it’s 20% more profitable, the entire industry will adopt it either way. 

Regardless of public sentiment, and often in opposition to it, institutional support drives technology forward in this way.

Taking the Hype Out of the Hype Cycle

The problems with the “hype cycle” actually begin with the name itself. At the risk of getting a little dry and philosophical, I’d like to discuss those problems in a little more detail. 

The Hype Cycle’s sin is one of incommensurability and category error. That is: it lacks any “common measure” for tracking the magnitude of the “hype” surrounding a particular technology, and it covers for this lack with a sleight of hand, conflating “Hype” and “Productivity” as if they belong to a single category of phenomena that can be meaningfully compared or combined. The term being measured on the left knee of the S-curve (the huge bump in public expectations) is fundamentally qualitative and subjective; while the knee to its right is quantitative and objective, yet they’re treated as one and the same thing.

It’s as if hype is just a kind of productivity, or productivity, a kind of hype. That is obviously silly. In this way, the Gartner graph “borrows,” in a sense, the credibility of a quantitative fact about a technology’s eventual real adoption and economic potential, and conflates that potential with the mere perception of potential represented by its hype.

Any clever marketer (and there are many) can retrofit a narrative about current productivity gains to an S-curve that suggests this gain is the product of a hype cycle. Post-hoc rationalizations have become something of a software industry obsession really, which is interesting considering it’s an industry so driven by the science of computation and data analytics.

Disconformity

This problem highlights the weaknesses of viewing technology adoption through the lens of economic impact alone. First of all, there is little economic impact when a technology isn’t widely adopted, so measuring the productivity it represents doesn’t make sense until that productivity appears. Secondly, perception doesn’t play much of a role in that impact if and when it does occur. Public perception less of an impact still.

There is thus no way to say that a technology has reached a “mature” level of acknowledgment in the popular media or public imagination. Many technologies simply become a part of the landscape, continuing to steadily improve without drawing much attention. No throughline exists between what we expect from them, and what they may or may not ever deliver in productive gains. This is a basic disconformity between perception and economic impact that becomes very relevant when we dig into specific examples. 

We are not asked to consider something like HTML 5 along the lines of its public reputation, for example. It was simply an evolution of Hypertext that incorporated multimedia containers in a way that the newest technology could efficiently use. The result was highly relevant to the end-user, as it was the means by which the entire world was going to access multimedia, and yet it had no meaningful public hype “bump,” at the time of its introduction. Everyone now uses it, but only a very few even know what it is. 

Surely one can wonder where the boom and bust cycle was for the introduction of mRNA technology, as another very relevant example. It was simply not commercially viable nor known outside its field, up to the moment that the necessity for it changed that equation, and then it became both viable and heralded as transformative. The economic impact of that technology is almost incalculable, yet it’s hard to see any hype around it as proportional to that impact. Surely there had been some, but nothing on the scale one might expect if the Gartner Hype Cycle were really in effect.

If we are going to argue that this hype cycle is not only blending categories across “visibility” and “productivity,” but also blending the meaning of hype between engaged and unengaged audiences, then what if anything can we possibly be measuring?

Discontinuity

Hype itself is a gooey concept in any discussion of economics. Media coverage and popular understanding of a concept aren’t equivalent in any way to its economic value as a mature technology, and that eventual economic impact doesn’t necessarily come with a proportionate or predictable amount of hype

Even hype amongst professionals or investors is not always proportionate to a technology’s future role in the economy. Plenty of technologies become suddenly relevant after many years of quiet development. Others simply never do. Sometimes innovation just stops happening for an extended period. Few new things are really learned in that time, as fundamental research operates somewhere else. That discontinuity between the knees of the curve suggests it doesn’t really operate in any real way. 

Examples abound if we look for them. The YouTube channel Gaming Historian did a fascinating documentary on the development of motion control system U-Force, that began all the way back in the 1980s, before all but disappearing into the ether for nearly 20 years, reappearing only after home gaming had completely changed. Few if any companies or individuals continued to work on them for all that time.

Don't Touch: The Story of the U-Force

This story is not a rarity. And while it could certainly be argued that this is simply an example of a very early peak in hype, with the U-Force and PowerGlove duking it out for early adopters, followed by a very late boom in real productive gains with working gaming systems… well it’s worth asking what value that might have for understanding history. 

The discontinuity and disconformity we have addressed gives the lie to the idea that the economic value of a thing is predicted by public or even private interest, when it is first discovered or hypothesized.

If that were the case, then the Nobel Prize shared by the Katalin Karikó and Drew Weissman in 2023 fundamental discoveries behind mRNA vaccines would likely have been awarded decades ago, when they were first made and published. Certainly inventors behind patents that arose from motion control technologies in the 1980s, like Dave Capper and Jaron Lanier, are important as well. But they may be first among those to tell you that their importance has little to do with hype. 

By the way, Lanier once authored an essay, “One Half a Manifesto,” that rejects what he calls the “fetishism” of technological innovation, and argues that the idealization of computers as total solutions to human problems is, in itself, a hindrance to true innovations, and not fuel for them, as Gartner seems to suggest. Certainly his experience working on technologies like PowerGlove and other VR and MR technologies in the 80s and 90s helped form these views. Public perception, far from an ally, has often been the enemy of progress in these sectors. 

“Quick is Beautiful”

It is not even clear, come to this, whether the hype in the hype cycle is always a good thing, or whether it can’t sometimes play a role in dooming a technology from ever becoming as transformative as it might have otherwise been, in a more rational and less hype-driven scenario. We cannot know whether a broad-based expectation that mRNA vaccines would save the world from a global pandemic, for example, might have introduced factors into its development that would have impeded its progress.

The Mathematical Physicist Freeman Dyson famously said that “quick is beautiful.” That if a project or a goal is too time-consuming, too expensive, too broad in scope, or requires too many steps, then any organization that supports it can find itself “experiencing sensations of paralysis and demoralization.” Thus hype drives organizations of all scales to invest heavily in new technologies, but also precludes the possibility of pivoting away from those earlier investments if new ideas turn out to have more promise. 

Hype can also, perversely, introduce variables that confound progress entirely. This is the tendency that Arthur C. Clarke lampooned in the 1985 short story “Superiority,” which describes a future military making the same mistake over and over again: wasting its resources on new and superior weapons while its enemy relentlessly closes in. Clarke based the story loosely on the then-recent Nazi German campaign to bring Wunderwaffe, “wonder-weapons,” to bear on its strategically outmatched war effort. 

Dyson cited a different example: nuclear energy, which he suggested may have enjoyed far too much public enthusiasm far too soon, leading to a process of technological “lock-in” that made future developments and potential new discoveries harder to justify. The early public investments in nuclear power locked in any number of poor practices and design philosophies that led, eventually, to major safety risks and economic challenges that put a damper on enthusiasm for nuclear power.

Dyson argued that because the technologies of the future are always developed using the technologies of today, decisions made in different economic or political circumstances years or decades ago can drastically change the possibilities of the present. In this way, today’s pandemic-era expediencies may suddenly produce a major breakthrough, seemingly from nowhere, but that breakthrough may engender future challenges that can change the course of history.

Dyson described a “bandwagon effect” with nuclear power, which saw existing military applications serve as the basis for civilian nuclear power projects, making early investments in nuclear energy appear “cheaper” than they really were. The initial cost of fundamental research, design, and testing were born by the government (beginning with the Manhattan Project), so early commercial plants didn’t include many of these costs. Since part of the purpose of nuclear power development was to arm the military with weapons (and warships with the power-reactors that would keep them running), the government was willing to underwrite a large portion of the early costs of development. 

But that willingness was limited. When the military necessity for nuclear research and fuel production inevitably dropped off, the forward costs of both would have to be applied to any investments in civilian energy production. The initial burst of enthusiasm had yet more consequences. As the nuclear powers disarmed following the Cold War, the “Megatons for Megawatts” program of 1993-2013 converted hundreds of tons of fissile bomb material into low-enriched fuel for commercial energy use in both the USA and Russia. 

But in that process, the nuclear powers may also have helped ensure that other alternative fuel sources and the reactor designs that might use them became less economically viable as a result of more available uranium fuel. Who would invest in thorium reactors when uranium is cheap and plentiful?

The result was a far less promising long term economic model for nuclear power innovations, and the consequences of that early enthusiasm continue to play out. Today, despite major theoretical advancements, and despite the severe growing impacts of global climate change, reinvestment in nuclear has been surprisingly slow and has faced enormous PR challenges, partly due to how it was initially developed and popularized: first as a weapon of war, and then as a too-perfect solution to growing energy demand.

The Early Adopter Problem 

Now we are sufficiently grounded in the reasoning of why I think the “Early Adopter Problem” may be a real and seriously underestimated problem in the tech industry. It may, in fact, be at the core of a looming crisis.

While we are now being told with great regularity and ever-increasing stridency, that the revolution in technology that will be wrought by transformer-based automation will accelerate change in the greater economy in a way that is hard to fully understand, it’s also hard not to suspect that this is somewhere we’ve been before. Is it an overpromise, underdeliver scenario, like with VR/MR in the 1980s, or an over-investment under-development problem that will play out decades into the future? We can’t know. 

It would seem to me that in many respects, our industry has bought uncritically into the idea that once a thing is in the public imagination, that thing is 1) completely inevitable 2) an absolute good and 3) definitely something that will matter a great deal in the long run. From cryptocurrencies to GenAi, or AGI (Artificial General Intelligence), once the idea itself has gained purchase in the popular imagination, the development of this market and the underlying economic model is a foregone conclusion.

At the heart of this seems to be the Gartner Hype Cycle hypothesis, justifying any means by which the public imagination can be stoked, because the result should always be overwhelmingly positive on a large enough timescale. 

You may sense that a straw man is being constructed, and I assure you: that is exactly what this is. No one believes these things absolutely, nor does everyone share all these beliefs. Yet our industry certainly seems to act as if it does believe these things, and if these beliefs are not as common as they seem, the naysayers are doing a good job of blending in. So if we are all acting pretty much as if the hype cycle is destiny, I will treat us as if we really believe it.

“The Hype Must Cycle”

As we’ve seen I think repeatedly through this post, 1) is demonstrably false. Plenty of technology developments never materialize because they’re too expensive, they’re too difficult, they’re in some way taboo or politically unacceptable, or they’re just plain good old-fashioned impossible. I will go out on a limb and say, for example, that I doubt a global decentralized currency with the steadfast value of precious metals and the safety and ease of use of cash is, even if technically possible, politically unacceptable to pretty much any developed nation. 

GenAi is getting pretty good at fooling people into thinking it can do more than it can, but we long ago went out on a limb to predict that promises of AGI coming from the tech industry are overblown, at best. They may be fundamentally flawed, and AGI, or a truly autonomous, self-aware, and sentient computer mind may well be impossible - perhaps one reason that OpenAi founder Sam Altman has been shifting the goalposts on what can be labeled “AGI.” The hype must cycle. 

Even if AGI does turn out to be possible and it arrives, as it has been promised, within the next year or two (mark that calendar), it will have arrived after an extraordinarily resource-intensive campaign of spending on infrastructure and energy that was by no means an inevitability. If AGI does arrive,  it will arrive as the result of a massive effort for an uncertain reward. It will certainly be one of the riskiest and most successful strategic investments private industry has ever made. If it happens. But even if it does, Dyson’s points should make us wary of insisting that 2) this outcome is an absolute good. 

There is good in most technologies (even Polonium, apparently), but there is also room to question how much. No transformative technology has ever been entirely to the good, and how much good a technology does is certainly something history eventually decides. 

Finally 3, as we’ve seen, is also simply untrue. Plenty of technologies deliver exactly what was promised by the hype, but turn out not to matter very much, or to be all that transformative in the long run. Which is fine, really. There’s nothing wrong with discovering that a shiny new toy is really just the same old thing in new wrapping paper. That’s progress, of a kind. 

“Buy the news, sell the rumor,” as they say on Wall Street. Or, hype is fine, but don’t expect it to always be followed by something real.