The only way we can still be saved from being turned into paperclip

Collin Staleph, visionary entrepreneur who changed the world, passes away at only 38.

It’s not without shock and grief that we report the death of the genius that transformed society. It’s hard nowadays to imagine him needing introducing, but let us not forget that it was not always like that. We felt like our best homage would be to remember how far he has come.

Collin came from very humble beginnings. Not much is known about his childhood. He studied Computer Science and Cognitive Science in France. He was a pretty average student, and his youth was mostly unremarkable. Nothing is worth mentioning besides occasional participation in activist movements and a few contributions to the “open-source” community under the pseudonym “k0l1nn”.

He really entered the public stage in the late 2010s with the creation of his first and unique company. The “Collin Crates”, as it was called back then, started like any other lootcrate service. It was a very popular trend of this time, propelled notably by massive advertisement campaigns on YouTube and social media platforms. From the famous “Dollar Shave Club” to novelty pop culture figurines or even movies, it seemed that there was a subscription service for everything.

The success of this model was understandable: who wouldn’t like to receive an unexpected surprise in the mail for a small fee? It was Christmas every month! Loot crates escaped the ethical debates about their virtual counterparts (i.e. “lootboxes”) by guaranteeing a physical object of a predetermined value. Furthermore, it was the time of the big explosion of the subscription economy, when platforms like Disney+ came on the scene to reshape the internet into the famous controversial model of American cable TV.

Although the future of this market seemed bright, this wasn’t without its challenges. In an already saturated domain, how could Collin compete with established giants? He surfed another trend of the 2010s era: machine learning.

All loot crates services claimed to be somewhat customized, but none of them actually delivered. From the start, “Collin Crates” wanted to be different. They wouldn’t focus on a given product like men’s hygiene or multimedia. Instead, they subjected each customer to thorough (but voluntary) questionnaires and fed the answer to a machine learning system which delivered a suggestion that would perfectly fit the tastes and needs of the client. That way, it was a surprise for everyone, but it was guaranteed to please (most of the time, of course).

The idea was pretty appealing, but the logistics were obviously challenging. Some may still remember the struggles to cope up with the hype in the beginnings. The service had to be limited to be invite-only, while Collin and his team worked tirelessly to scale up the infrastructure. Fortunately, economies of scale quickly came into play, and pretty soon the more customers they had the easiest it was to provide the goods.

It could have stopped there, as a one-hit-wonder success story of a novelty platform. The model wasn’t very durable. They operated on a very thin profit margin. It more or less amounted to a low gain dropshipping platform. That was without counting on Collin’s fascination for the algorithm.

The next breakthrough came from looking at the machine learning’s output. It was far from being perfect, and frequently underperformed. One of the problems the team struggled with was the fact that the system kept recommending basic necessities (food, groceries…). It was pretty understandable, considering it’s what all humans need most, but they had installed ad hoc filters to limit the crates to leisure products. Their strike of genius was to simply ask “Why not?”. Why shouldn’t they propose basic necessities to their customers?

That was the start of the rebranding. The “KolKrates” as we know them were born. The whole subscription pricing model was reworked: instead of a fixed price, people would now pay what they wanted and get a crate of equal value in return. They would simply select the categories of KolKrates they wanted (“basic necessities”, “superfluous leisures”, etc…) and the recommendation engine would simply do the rest.

This model was a resounding success. Who wouldn’t want to delegate their groceries to someone who could do it better than they could, who could take into consideration everything from bulk discounts to nutritious value or even ethical positions of brands (a big social issue at the time). The time for suboptimality was over.

From an operational standpoint, more customers meant more money in the system, and better allocation, planning and ultimately savings. The algorithm could factor in availability in its assignment of resources. For instance, it could grant people their second choice to prevent a shortage.

In addition to data about stocks, a new input for the system was the use of their new crawling technologies, which would gather all sorts of information from social networks of volunteers to improve its model of their preferences. Concerns for privacy were quickly outweighed by the gains, as people found themselves discovering new dishes or clothes they didn’t know they would adore. But the algorithm did.

Within a few years, a third of the population was subscribed to KolKrates, at various levels of commitment. There were already power users granting a wide part of their salaries to KolKrates, which managed most of their lives for the better. In a word submerged by an overabundance of choice, lessening the cognitive load and guaranteeing optimality were much appreciated.

It wasn’t long before KolKrates expanded its activities to consulting. Its massive database made it an amazing candidate for financial investment management, of course, but they were kind enough to create a completely free tier where the algorithm would share its insights and provide people with advice for all sorts of requests they might have, from choosing between purchases to choosing between careers.

It really opened the system to everyone. Anyone could try it, and few wanted to stop after getting a taste of its results. Soon, most of the population subscribed to the platform. Some people still shopped by themselves, but the efficiency of the algorithm was slowly winning over the few last remaining detractors. It wasn’t long before Kolkrates supervised the whole country, with its customers’ blessing.

Leveraging that influence, the company could make the world a better place. Following up on their motto to “destroy inefficiency”, it put an end to corruption and speculation. The resulting savings and smart allocation of the collective subscription money allowed the basic necessities of everyone in the system to be met. The surplus was shared from each according to their contributions, to each according to their needs, in the best possible way.

“But what is the best possible way?”, thus starts the suicide note that Collin left as he departed. “Giving people what they want is easy, until someone’s desires conflict with what they need, or with what others want. We did not solve inequalities and all the problems of the capitalist market. We simply replaced them with the question of the alignment of our AI. Of course it’s better than maximizing an arbitrary notion of profit. But choosing what to optimize for between people’s needs, wants, or happiness is a burden no man can bear. And I am not a god.”

The final words of our hero only gives us a tiny glance at the ethical dilemma he must have lived with everyday. It’s almost understandable that under this crushing pressure, he ended up taking his own life. More than anyone else, he had the weight of the world on his shoulders. In the end, even the KolKrate algorithm couldn’t lift it off from him.

Who knows what the future holds for the ethical alignment of the KolKrate AI, the aptly named “Maximal Alignment Resources X-changer”? As of now, no one can tell. The only “god” it responds to is the crowd. But one thing is for sure, it will keep using the funds of its voluntary subscribers to maximize their satisfaction, and not simply shareholder profits.