The last time I posted on this blog it could best be described as a lament of the death of the AGORA Initiative of the ULISSES Project at the University of Utah. To be clear, I did not see much of a reason to post on this blog if the AGORA Initiative was dead, and for the last three years it has effectively been in a coma, as the Initiative was only surviving in the work of my former students who were faithfully applying the applied methods we developed at the University of Utah with their own work and students.

Today, I again pick up the pen (or more literally keyboard) and state that while the AGORA Initiative did die at the University of Utah, it was reborn in the AI Projects of my students and their students. The very Jupyter Notebooks which I previously stated were going to be a “textbook,” were used as the foundational files for AI work. It became so much more and evolved so much faster than I could have imagined, mirroring the developments that we have seen in the past three years in AI. Because of this, as I restart this blog, I feel it necessary to return back to the purpose of the ULISSES Project. This restarting of the ULISSES Project blog also happens to correspond to my CEO asking me to start a blog on our work commercializing the technologies at Flex Axion Enterprises. For sake of continuity my historical blog posts on the ULISSESProject.org will be replicated on the related Flex Axion blog.

Over the next three days, I’m going to articulate this “homecoming” in three blog posts, one dedicated to each of the three initiatives of the ULISSES Project, and articulating how they relate the current AI Revolution, and to our company, Flex Axion Enterprises. I start with AIAX, and then finish with the reimagination of the AGORA Initiative that will be launched this later this week.

The AIAX Initiative – The Ontological Underpining

Taking a 40,000 foot view of the ULISSES Project, the AIAX Initiative had always been the core of my research going all the way back to graduate school. In short, the AIAX Initiative is an ontological framing mechanism that allowed me to harmonize the models of the complex world of finance. It was when working with an emerging markets hedge fund in New York with Bloomberg and IBM that I found that AIAX was just where the complexity begins. You see, I once I was able to code up my complex models completely, I found that the processing costs would dwarf the gains made by the associated trades, and that was “IF.” I say a “a big IF” because I also found that we could not do the processing in time to trade. At that time, I was told by the man running the financial commercialization of Watson for finance for IBM that they’d provide me free compute to complete our project. Now to be blunt that made me laugh at the dinner table. Of course they would provide free compute for a while, but in hedge fund land if it doesn’t make money, you have to rethink what you’re doing or how you’re doing it. Simple calculations on my part showed that even with free compute, the methods we were using were fundamentally incomplete and had to be fully reimagined.

At the hedge fund we simplified and then simplified again and got great results, but that was from models that were millions of times more simplified that what I was able to code up. That fund worked wonderfully for three and a half years, and got a lot of attention. When the tech team was head hunted by a former Renaissance Technologies OG, rather than joining that foray, I stayed at my work that I’d already started at the Computational Finance Lab at the University of Utah, to fundamentally re-engineer how we address complex compute problems, something that required a much better understanding of how data relates to complex models.

So, what was AIAX? AIAX was and continues to be an ontological framing mechanism that allows me to coordinate complex financial models. My friend and mentor Barr Rosenberg harmonized these models before me with massive simplifying assumptions, the whole while using the collective works of Shakespeare as a metaphor to explain how he was able to “model the models.” He went on the record saying it was the “interplay of diverse interests enacted so eloquently by Shakespeare’s characters” that allowed him to imagine his system. For me, I built on Barr’s understanding (and even his coding techniques), and then used Dante as my driving metaphor. Some might think it is strange that we both used literally models as our “model of models,” but upon reflection that is how we as human beings think…we think in language and these language models (Shakespeare and Dante) provide two of humanity’s most complex literally metaphors that led to new maps of world knowledge.

Dante and his later interpreters such as Giambattista Vico argued that poetry serves as the unifying key linking literature, history, religion, philosophy, and politics. Vico’s work was an encyclopedic effort to join the arts and the sciences, providing a form of unity in stark contrast to the fragmentation of modernity, something that we are fighting against now. Long before people so casually spoke of “ontologies,” Barr and I were talking about the desperate need to merge systems of knowledge.

Akerlof, Morgan and Ontological Frameworks

Model integration was on my mind when I spent a week with Barr Rosenberg before I started this part of the ULISSES Project at the University of Utah (technically I’d started the ULISSES Project years before, but this was a new perspective and initiative). While Barr’s conversation started with Buddhist mandalas, it soon turned to computerized models and representations of reality. This started with George Akerlof and his idea that we as human beings seek to make sense of complex systems by using metaphors and stories as foundational frameworks. That then led us to Mary Morgan, the prominent philosopher of economics, who rather than viewing models as abstract mathematical equations or unrealistic simplifications of the real world, she views models as “working laboratories.” Ultimately, this idea coalesced into providing a consistent and coherent ontological layer so we could integrate models into a more complete simulation that could in turn give greater insights.

For students of Palantir, this laboratory of finance is precisely how they started at the hedge fund Clarium Capital. While Clarium went the way of the dodo after some of the biggest losses in the industry, Palantir was the residual. They then spent eight years trying to tame finance, and then only after that, turned to the government work. Make no mistake it is a difficult endeavor that one should not take on lightly. Thankfully it was work that I went to graduate school for, had worked on for years and it’s why I found myself talking to a former billionaire (he’d donated all his money to Buddhist causes) Buddhist monk who also happened to be one of the best financial modelers in history.

One of the things we spoke about way back then was one of his old mentors, Richard Mattessich, the creator of the modern spreadsheet who also is the person that in many ways presaged the ontological confusion we have in the world today. While Mattessich seemed so sure of himself back in 1964, by the time Barr and I were speaking fifty years later he proposed as his swan song a radical reimagining of the ontological landscape of economic and social sciences. Building upon his philosophical groundwork laid in 1964, he argued that reality is not flat or purely nominal, but is stratified into nested, emergent layers. This is precisely what we discussed, rejecting the “flatland” of neoclassical economics, something he referred to as Maya.

Beyond Flatland

This view that Barr and I shared is a perspective of a firm as an integrated whole. While neoclassical economics tended towards analysis that viewed a firm as a detached node in a supply chain or a simple profit-maximizing function, we were viewing things in a way more consistent with the Jesuit theologian and scientist Pierre Teilhard de Chardin, that companies and people are defined by their connections. simultaneously operating at different levels. This view, that information and accounting are structural models of a multi-dimensional relational reality became the basis from which I began my ontological work, knowing the end and then going back to the beginning. I use this perspective today to provide a generalized vision of the firm as a multi-layered entity where physical operations and informational ledgers are bound by specific mathematical relationships (such as clean-surplus accounting). By anchoring informational models directly in the various ontological strata of reality, we bridge the gap between between deductive mathematical rigor and inductive empirical observation.

So how does this relate to the current perspective of the world. To be clear, there is so much ontological confusion. Computer scientists have become obsessed with reduced models that work to a certain degree, but fail when they scale. This is precisely why I’m going to be relaunching the AGORA project this week–to show the world how a proper ontological model organizes information. The world needs to see an ontology put into play and that’s what we’ll be showing through the most tangible mechanism possible, real company results. Oh, by the way, those results will be brought to you by the technologies that we’ve developed at Flex Axion Enterprises.

It’s good to be back.