Why Everyone is Wrong About Google and the Innovator’s Dilemma

It’s fashionable to kick Google right now, but don’t count them out in the AI chat race

Folks lining up outside just to kick Google

While the hype around OpenAI’s ChatGPT has led many to declare Google as being doomed, this groupthink misses some key factors.

The theory of The Innovator’s Dilemma doesn’t neatly apply here.

The Dilemma in a Nutshell

Here are my notes taken from the book.

1. The best run companies are highly prone to disruption.
2. Good management listens to its customers in its current “value networks”, makes the improvements they want, kills off unprofitable projects, and focuses most energy on the most profitable endeavors.
3. This is all sustaining innovation; the current technology is improved over and over to sustain it’s life in it’s current markets.
4. Eventually, the technology overshoots the performance demands that the market desires.
5. This makes the company susceptible to disruption as the existing market is ripe for a shift in what it values in a product, from performance (and possibly weight, size etc.) to reliability and then cost as the product category becomes commoditized.
6. Disruption starts when existing technologies are redesigned or recombined in a new way that falls severely short of the current markets’ needs in some aspect, usually performance, but has other new characteristics, such as compact size, that are not valued by existing customers.
7. The company is unlikely to pursue the new type of product because its customers do not want it and it is not profitable.
8. Even when it does pursue it, it fails to excite people used to enormous sales contracts and management will take key people off the project.
9. Existing managers are less experienced in chasing and discovering new markets.
10. They listen too much to their existing customers who lead them astray.
11. Disruptive innovations must find new markets who can apply the technology in some way to solve a problem that’s unique to them, and they value the things that other customers do not seem to care to pay for.
12. This stage of product development requires keeping the product simple and cheaply adaptable, then iterating while watching exactly what customers do with the product, and not what they say. It also requires tolerating frequent failure and doing so without spending much money. These are skills for a different set of people.
13. The technology finds a beachhead market and sees some success. It begins to improve along a dimension, such as performance, at a rate faster than the rate of improvement desired by the customers of an adjacent market such that at some point in the future, the disruptive product pushes “up” and into those markets and takes over.
14. Otherwise, the new kind of products that the disruptive technology enable, become the new normal and the old products disappear; think small hard disks and laptops eating at sales of desktops and people using a laptop fulltime.

The Innovator’s Dilemma, as outlined by Clayton Christensen, describes how incumbent companies struggle to adopt disruptive innovations that initially underperform on mainstream attributes and cannibalize existing product lines.

But aspects like points 6–11 about disruptive technologies lacking in performance and struggling to gain traction within the incumbent don’t hold true for Google and AI chatbots.

• Doesn’t fall short, like how phone camera sensors were shit at first.
• Customers want it.
• There are no sales contracts, they not making parts, and the sales of AI generated adverts will be very lucrative and exciting.
• This is not a new market.
• They have been led astray by their customers but not in the way it means in the book. I’ll explain below.
• They not looking for a new market for an intially-shit product.

Large language models (LLMs) offering question-answering and dialogue aren’t lacking in performance — if anything they are amazingly capable compared to traditional search. The ability to get succinct, human-like answers rather than sorting through webpage links is hugely compelling for users. Google’s own employees and management should be excited, not dismissive, of this product direction.

Moreover, Google has been at the forefront of LLM research for years, with their work heavily relied upon by OpenAI and others. For years now, they’ve been incorporating LLM-generated answers into their search results. Transitioning more fully to a chatbot interface would be an evolution facilitated by their deep AI expertise, computing infrastructure, and data resources — not a disruption underperforming in key metrics.

The profit model concern also seems misplaced. While LLM chat may reduce the need for URL link listings, Google is superbly positioned to monetize it via advertisements — especially highly contextual and dynamically generated ads customized for the user and conversation. Their dominance in digital advertising gives them an edge over cash-poor startups.

Cost of operating LLMs at scale is certainly high, but Google’s operational cash flow and cloud infrastructure put them in a far better position than cash-poor upstarts. Moreover, Google can likely afford to provide its best LLM models for free search — which is what billions of users are accustomed to and expect.

While the tech pundits and early adopters on Twitter may flock to paid services like ChatGPT, the vast mainstream user base has Google cemented into their habits and frontal lobes. There are also plenty of use cases, like researching niche topics in-depth or finding the original source websites, where searching the open web will remain necessary rather than relying solely on a closed LLM’s summary.

So while ChatGPT was an impressive feat by OpenAI, suggestions that Google is doomed to disruption seem premature at best. With their talent, infrastructure, and profit model advantages, they have strong positioning to remain highly relevant as AI interfaces evolve. Writing them off would be unwise. The disruptive innovator’s dilemma does not neatly apply here.

However, it’s undeniable that ambient computing interfaces embedded into homes, cars, and devices pose an threatening transition that even Google’s services may struggle with long-term.

While Google has made inroads into some of these surfaces like mobile OS and home assistants, they’ve also lost their way at times. Exemplified by blunders like the Damore firing and being bullied into backing down on forward-looking products like Google Glass due to public backlash.

Regaining that innovative edge will be critical as we shift to an AI-first world saturated with intelligent data streams surrounding us.

When ads and links come to ChatGPT or Claude, that’s the time to worry, because that’ll provide the cash to provide the highest intelligence service for free.