This month’s Frame: rethinking competition in the AI race
A framework for long-term collective success.
“Carthago Delenda Est.”
This cryptic phrase, emblazoned in Gothic script on Mark Zuckerberg’s t-shirt at his 40th birthday party, quickly caught Silicon Valley’s attention. “Carthage must be destroyed!”—was originally the rallying cry of Roman senator Cato the Elder, who ended all his speeches with an exhortation to push for war against Carthage.
In Facebook’s early days, Zuckerberg adopted this motto to remind employees that he viewed the tech landscape as a battlefield, with Google as the prime adversary, especially during the launch of Google+, a (short-lived) social network which threatened Facebook’s dominance.
It is no coincidence that the Meta CEO resurfaced this martial motto. The difference is that the competition between the tech behemoths isn't about racing to grow social networks, it's about foundational AI models, with Open AI / Microsoft, Google and Meta out in front.
Despite this backdrop of heated competition, cage fight challenges, and clear differences in approach (eg. open source versus proprietary) from both a systemic and an individual company perspective, the best way forward might be less competition and more collaboration.
The most well-known problem of game theory, the prisoners' dilemma, helps clarify why this might be the case.
The framework
The Prisoners' Dilemma, a fundamental concept in game theory, was formulated in 1950 at the RAND Corporation by mathematicians Merrill Flood and Melvin Dresher. Albert W. Tucker later popularized the concept.
The dilemma involves two prisoners who must decide independently whether to cooperate (remain silent) or defect (confess) without knowing the other’s choice. The outcomes are:
Both cooperate: each receives a light sentence.
One defects, one cooperates: the defector goes free, the cooperator receives a heavy sentence.
Both defect: each receives a moderate sentence.
Most people know that the learning of the experiment is that, since they don't know what the other player will do, if each participant is rational they will defect. This leads to a systemic equilibrium characterised by competition, rather than cooperation.
This model seems to describe the behaviours of tech companies developing AI models aptly: each worried that the others might beat them to market, they are racing to ship products as fast as possible, even if it is at the cost of suboptimal user experience or lack of ethical guidelines.
However, a variation of the Prisoner’s Dilemma, introduced by political scientist Robert Axelrod in the 1980s, changes the dynamic. In his version, the same participants play multiple rounds, with each round’s decisions influencing future behavior.
To know which strategy is optimal in this more complex situation, Axelrod organised an open competition, where academics submitted their strategies. Computer simulations were then run to determine the winner. Much to the participants' surprise, the winning strategy was deceptively simple: in the first round, cooperate. In the following rounds, mimic the other player's strategy from the previous round: cooperate if they cooperate, defect if they defect (the strategy became known as "tit for tat").
The findings from Axelrod's experiment are profound: as soon as you add the time dimension—which is inherent to most real-life decisions—the optimal strategy is not selfish but cooperative.
Using the framework
The tech industry’s focus on rapid innovation often emphasises short-term gains over long-term trust. Incorporating a temporal dimension into strategic thinking can foster cooperation, leading to more sustainable, ethical and ultimately, more successful business practices.
These learnings seem particularly urgent for social media platforms. Much has been written about the "enshittification" of the internet, with users and pundits remarking that both the content and user experience of online platforms are deteriorating. Short-term profit motives, such as prioritising extreme content to drive clicks, often drive this decline.
Yet as Axelrod demonstrated, developing a relationship based on mutual trust is essential for maximising benefits over the long term. If users believe—as they appear increasingly to do—that their relationship with tech companies is a zero-sum game, they will leave for better alternatives as soon as they arise.
One promising product strategy avenue to explore for social media platforms is interoperability: instead of trying to lock users in a closed ecosystem, make it easy for them to port their data between different platforms. The resulting ecosystem could be a healthy mix of competition (each platform trying to create the best user experience) and cooperation (each platform and user benefiting from a common tech foundation and a fungible user base).
It is worth remembering that the web's foundational protocols (eg. HTTP, DNS, TCP/IP) were established through mutual agreement among early participants in the development of the internet. However, as new technological paradigms emerged (then social networks, now artificial intelligence) it has become increasingly difficult to envision current tech giants cooperating to define common protocols and ethical guidelines.
Just as Rome's victory over Carthage did not lead to sustainable peace but to a series of further conflicts, the tech industry's relentless competition (and sometimes short-termism) could lead to a similarly unstable and fragmented future.
The main learning from the Prisoners' Dilemma is that trust—with users and between competing companies—is earned over time. It is therefore by embracing interoperability and focusing on creating shared standards and ethical guidelines, that tech companies can ensure their sustained collective dominance, all the while building systems and experiences that deliver against the needs and expectations of their users.
News
We're delighted that Stripe Partners will be making five presentations at EPIC, the premier conference for ethnography in industry in Los Angeles, 18-21 August. These include cutting edge integrated social and data science work conducted with Google, an exploration of the subscription economy, and a new framework for understanding how teens manage context collapse on social media. If you're attending come and meet us. There are still tickets available, get yours now.
Frames is a monthly newsletter that sheds light on the most important issues concerning business and technology.
Stripe Partners helps technology-led businesses invent better futures. To discuss how we can help you, get in touch.
