6 minute read


In this post, I revisit the question Tim Wu raised 13 years ago about the Internet: are dominant IT entities today distinguishable from their predecessors who followed a common pattern of rise and fall? Rather than listing and weighing cases for and against the question, I focus on related patterns we see from the public and the companies that will likely persist in the current context of noticeable advancements in AI.

Is the Internet different? Written in the context of the ever-expanding role of the Internet, Tim Wu’s historical analysis of monumental shifts in the information technology industries addresses whether the Internet will stand apart from its predecessors - telephone, radio, TV, and cable - in defying the precedents set by the latter. In particular, the cycle below summarizes the rise and fall of the titans that held near-exclusive power over the defining technologies of their time:

A new technology with the potential to disrupt or destroy the existing dominant technology arises from a small group of inventors/hobbyists. At inception, this novel technology generally maintains a freely accessible channel that allows quick adoption, hence expansion. Then, often with the help of deep-pocketed investors and the state, the leading entities replace the incumbents and consolidate their economic clouts. Over time, they become industrial behemoths resembling, if not outright, monopolies that oversee once an open system, which has now become closed where the flow and nature of content are “strictly controlled for reasons of commerce” (p. 6). Such a shift hinders innovation as these giants are more incentivized to protect their established technology generating tremendous wealth than to develop something that could replace it. Ironically, reaching this profitable yet stagnant stage inevitably reveals the limitations of the technology and creates an opening for the new entrant, whence the new cycle begins.

A little more than a decade has passed since the book was published, and the author’s question seems current, especially in the backdrop of incessant posts and headlines about AI. Whether the Internet is different or not, I believe that there are certain implications we can derive from historical patterns described in the book. This is, in a way, to be more robust; except for lucky and exceedingly hardworking few, foreseeing technological advances even in the near future is a daunting task for us amateurs. Therefore, holding a mindset that can register and adapt to what takes place could be a more promising bet. Below is my attempt at that.

We Who Lack Imagination

This is not meant to be derogatory or condescending, as I am certainly one of many who are content with existing services and products from leading tech companies. One recurring pattern on the part of the mass public is that we simply lack the imagination to “actively demand” change or innovation. Put another way, we are generally not aware of the drawbacks of existing technologies until a new invention that makes them salient catches our attention. Indeed, the long delay and the limit in the message length of telegraphs became obvious only in comparison to the telephones. Another important point is that the public might be slow in noticing issues with existing technologies, but they are not as slow in adopting new technologies. (Just consider the speed at which consumers are abandoning cable TVs for streaming services today.)

So what does this imply? The seemingly invincible status enjoyed by tech giants today (e.g., the magnificent seven – Apple, Amazon, Alphabet, Nvidia, Meta, Microsoft, and Tesla) may not be so after all. Take Apple as an example: just this year, about 50% of its revenue will be from selling iPhones alone. Suppose a new communication device comes along and dethrones smartphones as our main communication mode.1 Surely, Apple has more than enough to quickly launch a device with state-of-the-art technology, but we have seen what happens to second-movers (e.g., Nokia and BlackBerry).

Conservative Incumbents

Moving onto the companies, one notable issue with the status of near-monopoly pertains to the change in the risk preference of the organization as a whole. More specifically, the immense fortune from its existing technology renders a “conservative” approach of selecting improvement over replacement with disruptive technologies more sensible for the incumbent. The story of Bell Labs and its early development of the answering machine is the case in point. An engineer named Clarence Hickman developed the answering machine in 1934, but the senior management of both Bell Labs and AT&T blocked its productization on the grounds that it would “greatly restrict the use of the telephone,” the company’s main revenue source. Wu suggests that this epitomizes the weakness of a centralized (leading incumbent-driven) approach to innovation:

AT&T, as an innovator, bore a serious genetic flaw: it could not originate technologies that might threaten the Bell system… the Bell Labs was practically restricted to sustaining inventions; disruptive technologies, those that might even cast a shadow of uncertainty over the business model, were simply out of the question.

Tim Wu, The Master Switch, p. 107

This flaw was rather costly, as it would take another 17 years for the answering machine to reach consumers.

Such a pattern does not seem unique to the early 20th century. Sundar Pichai, the CEO of Alphabet, admits in his interview at Stanford GSB that “companies become more conservative as they grow.” A recent internal leak from a Google engineer notes that both Google and OpenAI are falling behind open-source developers in developing AI models, which is attributed not to the lack of talent or capacity but to structural reasons (i.e., bureaucracy). As shown in the answering phone episode, this has notable implications for consumer welfare: the dominant industries might be (already) hurting the consumers by hindering the development of the disruptive technologies that could have blossomed quicker had they not taken such a defensive stance. In the context of AI and its concrete harmful cases from misuse, taking a slower route to innovation may not be a disservice to society after all. Nevertheless, the tendency for the dominant entities to become more conservative in size remains, providing a case to be more cautious when these companies call for more regulations.


In sum, I delved into recurring patterns on the consumer and the industry sides to arrive at implications that would hold whether the Internet is truly different or not. Given the noticeable difference in the demand and the adoption of new technologies by the public, dominant conglomerates, however implausible as it may appear now, could come and go. Moreover, the propensity for the leading incumbents to be conservative should prompt us to question the alignment of interests between them and the public. While it would be easier to simply accept the status quo as is, too much is at stake to let these companies blindly have their ways. I leave readers with the author’s answer to the opening question:

There is a dark underbelly to the diversity of content and services that the Internet has brought us, one that leaves it more vulnerable to centralization, not less…The prospect of a new imperial age, even if only partially visible now, seems to me as likely as it ever has been at this point in the Cycle. This time *is* different: with everything on one network, the potential power to control is so much greater.

Tim Wu, The Master Switch, p. 318

  1. For example, take a look at the new device from Humane, a company founded by ex-Apple engineers.