Intellectual Scouting
The Archetypal Scout
Some time ago I read an autobiography by Major Frederick Russel Burnham about his time scouting on the American frontier and in present day Zimbabwe. Reading this book brought about a fascination with another person I have rarely had. This was a real person, living and thriving in the wilderness and exploring it throughout his life. A person whose influence is felt in even modern day scouting movements (through his correspondence with its founder Baden Powell). More importantly, however, was my realization of the significance of the formal occupation of being a scout.
Throughout history, armies have been most dependent on scouting parties for their very survival. The skill and ability of its scouting parties were the direct determinant of its ability to succeed. The goal has almost always been to go into the wilderness or enemy territory and derive structured and valuable information to be used for strategic positioning.
This has been the case for Major Burnham, both on the war front but also for mining. In the realm of scientific discovery, scouting expeditions such as those of Crocket and Boone, Shackleton and even the moon landing were all great efforts of going into the wilderness and capturing valuable information from it.
Disappearance of the Physical Scout
To a large extent, the formal job of scouting for war, scientific or industrial reasons has largely disappeared or at least shifted. Modern war employs drones and spyware to get such information. Exploring space and deep oceans are mostly done by telescopes and robots. Technology in that sense extends our ability to discover novelty in nature.
It would almost be tempting to believe that scouting itself has died out completely (other than the formal intelligence agencies). But the archetype is also very much alive. We could define a scout as one who goes into the unknown, observes and analyses information, and then uses this information for improved decision making. The idea hasn’t died out at all. It just moved to an intellectual stage.
Intellectual Scouting
This movement to an intellectual stage needs to be consciously and formally acknowledged, though. I call this process intellectual scouting: the deliberate discovery and structuring of information that is not yet widely available. The world we live in is highly digitized yet much of its information is completely uncaptured.
The internet is made valuable by what is called the wisdom of the crowds. Everyone with something to share can do so and others can interact with it. Unfortunately, our brains aren’t very good at taking enormous amounts of information and structurally synthesizing so that that information can be used for any one tasks it is presented with. I shouldn’t say it’s bad at this necessarily, it’s actually very good at synthesizing. The enormous amount of information available just makes it infeasible to scale this up. The rise of large language models on the other hand directly solves this synthesizing issue and makes it perfectly usable for any one person. We just have to interact with a model.
But consider the information scarcity problem. These models are trained on existing data, much of it sourced from the internet, yet their widespread use reduces the natural incentive for individuals to contribute new knowledge. As more content is generated by AI itself, there is a risk that future models are trained increasingly on recycled outputs rather than on original human insight.
Being able to go into the unknown, collect and analyze data and then bring it back to improve decision making becomes quite valuable in this context.
Examples
A clear example of intellectual scouting would be that of the scientific researchers. They develop hypotheses and test them through observation and experimentation, in turn producing new knowledge about the world. This is a form of inductive reasoning, moving from specific observations to more general conclusions.
Intellectual scouting isn’t limited to formal science, however, and can be used in all kinds of cases. It could take the form of structuring information that already exists, but hasn’t been able to be captured. The world is extremely complex, and on every plane of abstraction there are phenomena that produce valuable information that if captured, would lead us to more informed and strategic decision making. By recording and organizing the world directly around us, but also inside us, we produce data no one else has access to.
We could create genuine value by producing information instead of just consuming it. And this I posit to be turned actionable through the idea of Princeps or Principal AI.
From Agents to Principals
What can be done with this? Well that’s a difficult question to tackle. One possible way is for companies, governments or institutions on the one hand and individuals themselves on the other to start recognizing the value of Intellectual Scouting. Real importance should be given to discovering new information that isn’t typically available on the internet, or whose capturing isn’t typically made.
This then leads us to the crucial point that private information is private for a reason. We can view ourselves as competitors in a marketplace who must defend their proprietary knowledge because if it became public there’d be no competition. Without competition, there is no income for anyone, nor is there innovation and welfare for everyone. The concept of intellectual scouting can only be applied when we act in accordance with this principle of marketplace participants, perhaps more accurately termed economic agents.
Now we’ve all been taught about the principal-agent theory or problem. The idea is that different incentives between the principal and the agent leads to structural friction. The principal owns the entity and the agent is employed for their specialized skill in order to further the interests of the entity. It can be reasoned that agents may have different incentives than the principals, namely to seek their personal maximal welfare instead of the maximal welfare of the entity. In companies, this plays out between shareholders and managers, and incentive structures are set up to mitigate this and align the interests of the managers with those of the principals.
When we see ourselves (person, entity, business, institution) as economic agents in a marketplace with proprietary knowledge to protect and gather, we often are both principal and agent. But when we employ AI models such as Claude, Gemini, GPT, and others for executing tasks and furthering our interests, we separate the agent from the principal. This is what we see when we personally use such models for making decisions (developing ideas, reflecting on decisions, gathering information) or when a company uses RAG (retrieval augmented generation) where it sends their proprietary documents together with a related question to such model for quickly gathering information.
We essentially turn our knowledge generating system into something resembling the consulting relationship. The problem with consulting in this framework is exactly that of the principal-agent theory. The agent (model) is incentivized to ensure its own continuity and generally doesn’t have aligned interests with the principal (economic agent - person, business). In business, this is often mitigated by contract law and reputational capital, but the underlying tension remains. Informally this is the exact reason Machiavelli advised against the use of mercenaries.
What to do about this and provide a playing field for intellectual scouting to deploy naturally then? An idea here might be to move from models as agents to models as part of the principal unit. The idea is similar to the RAG described above, but instead we take an open source model that comprises all the available information available on the internet, and then tune it to be personalized and make us of our personal information.
The idea of intellectual scouting can then be discussed on two levels:
Institutions: Here, companies and other institutions might increase competitive advantage by utilizing not only all available information, but by actively creating and gathering information and incorporating this into their models.
Individuals: Similarly the idea can be proposed to increase individual competitive advantage by increasing the information used in our lives. For instance: we go to the store for groceries on a weekly basis. During these visits we often buy the exact same things, but never optimize the information that’s there. We buy two packages because having a shortage in the middle of the week is annoying. But do we actually use two packages? Have you actually ever tracked your own consumption habits to verify whether you buying habits make sense? This is the type of information generation I’m talking about.
By using our personal ingenuity to uncover and structure information that no one else has thought to utilize, we create proprietary information. By combining this unique information we personally scouted out (Intellectual Scouting) with the already available pool of synthesized information available, we can further our personal competitive advantage.
The major reason such intellectual scouting is valuable is that no one really does this. Scouting is not a topic formally recognized as being valuable in the area of intellectual property. In a business context, all information is automatically reasoned about in a commercializing or strategic manner. On a personal private level, however, how many people do you know who actively seek to maximize the benefit of each piece of albeit valuable knowledge or information that they own? By finding ways to turn everyday actions into useable information to train such a model on, a personal model can be developed on top of the general one seen in GPT that is able to work for you in a valuable manner.
I’d call such a model a principal model and further extend the terminology to PrinAI (Principal AI). We then have a nice counter balance to the (un)natural principal-agent problem we face on a societal level.
Synthesis
The modern world has more data available than ever, yet somehow there is scarcity of real and original signal data. In this context, the archetypal scouting ideal that seemed to have lost its footing over the past century now regains real importance and value. By employing intellectual scouting we are able to uncover strong original data that prior technological advances haven’t yet used to their advantage.
Combining such newly structured data (call it proprietary knowledge) with already available pooled general information can give us truly newfound competitive advantage, both on a personal and business level.