Human Frailty: Why Superset Exists
--
I’ve always been an optimist — especially when it comes to people. There’s a verse from a Rumi poem inked on my forearm that reminds me as much,
“Out beyond ideas of wrongdoing and rightdoing, there is a field. I’ll meet you there.”
Despite my optimism, I understand that human beings are frail. To be human is to make mistakes and, I would posit, to rise above them in the next moment. Perhaps not coincidentally, the origins of Delphia can be found in the ashes of Cambridge Analytica.
On a Monday in late October 2014, I was introduced to Dr. Clifton van der Linden, a scientist whose novel approach to modeling public opinion required the trust and cooperation of the public itself. He and his team had built a tool, Vote Compass, which helped people discover their alignment to the political candidates running in a given election. To generate this insight, participants needed to answer dozens of questions about their values, attitudes, beliefs, and ideologies. The lab would then use this data, in combination with the social media behavior displayed by participants, to forecast elections worldwide with unprecedented accuracy.
We were introduced because “Cliff” was struggling to find a sustainable business model and a mutual friend thought I could help. The obvious model was advertising — many political parties had approached him about accessing the data, but each time, Cliff turned them down. He knew the data would ultimately be used to influence people nefariously and, in his eyes, it wasn’t worth the millions of dollars on offer. I admired that.
Coincidentally, the same year I met Cliff, Cambridge Analytica made a name for itself by influencing 44 U.S. political races using similarly novel methods of profiling the general public. We all know how that ended.
I spent the next three years donating my time to Cliff. I wanted his lab to succeed as a business, but more than that, I wanted his principles to succeed. Chief among them was that if you can work in partnership with people — and genuinely represent their interests — a whole new world of prediction was possible (and perhaps Google hadn’t won).
That felt like a powerful idea to me. One I could devote my life to. So I did.
Governance As An Experiment
Elsewhere on the internet in 2014, another revolution was just getting off the ground. It was in May of that year that Vitalik Buterin (re)-popularized the idea of decentralized autonomous organizations (DAOs).
The basic premise of a DAO is that people can collaborate and govern themselves with code. In Vitalik’s words,
“The idea of a decentralized organization takes the same concept of an organization, and decentralizes it. Instead of a hierarchical structure managed by a set of humans interacting in person and controlling property via the legal system, a decentralized organization involves a set of humans interacting with each other according to a protocol specified in code, and enforced on the blockchain.”
In the intervening years, we watched the rise and fall of various DAOs in the name of experimentation. Along the way, we learned some important lessons. Most notably, that good governance is NP-hard, mostly for logistical reasons. Participants in a DAO (“the people”) don’t equally care about each decision, which leads to less than ideal engagement in the governing process itself. Decision-making complexity also grows non-linearly with each new member. When you layer immutability on top of these challenges, it almost becomes more trouble than it’s worth.
Unless the situation demands it.
In Delphia’s case, we are asking people to trust us with their data so that we can use it to improve their investment returns. We’ve seen this model work on Wall Street (hedge funds already buy our data without our permission) and we believe the same investment benefits should be made available to the average investor — especially since it’s their data to begin with.
When someone encounters Delphia in the wild, their decision about whether to trust us bottoms out on a few simple questions;
- Does Delphia have enough privacy and security in place for me?
- Has Delphia historically been a good steward of peoples’ data?
- Is the upside of sharing my data worth whatever risk exists?
- Who are the people in charge here and do I trust them?
It’s that last question that a DAO attempts to address.
While I might have conviction in my own moral code, I’ve always approached the building of Delphia as though it will outlive me. The further I push out in time (in my mind), the less certain I can be about who will ultimately have power and say over the data set we are amassing.
For that reason alone, it makes sense to have checks and balances. The first step toward that end is Superset — a DAO with a very important mission:
To ensure people receive a fair share of the value created from their data.
Superset is not owned by Delphia. It is run by a group of Trustees who’ve been given permission by Delphia to revoke the consent of any and all DAO members in a single shot. This mechanism ensures Superset can provide a check on Delphia’s power over its users. Superset board member, Amber Case, did a wonderful write-up detailing how this works in practice today.
Why Would Delphia Choose To Do This?
If your competitive advantage is trust, then you need to outperform the status quo. If the status quo are today’s terms’ of service, then it raises the question, “What would be 10x better?” One of Delphia’s founding engineers put it simply when we posed this question internally (emphasis my own),
“We want people to enjoy terms of service, not terms of servitude.”
Delphia is a bet that people will feel more confident using their full arsenal of assets to their build wealth if they know that an independent third party is empowered to advocate for, and enforce, their rights.
The other reason Delphia is choosing to pursue this is because I am the CEO and these are my beliefs;
Human beings are frail. We make mistakes. We err in our ways. It’s always been my opinion that no one person should ever govern something as important as our greatest collective asset — our data — no matter what they might stand for, or how pristine their track record. Things that are built to last must be resilient. Governance must be inclusive. And trust must be earned through the actions we take, not only the words we choose.
In essence, the goal isn’t, “don’t be evil” — the goal is, “can’t be evil”.
—Andrew
