- Unsupervised Learning
- Posts
- Desired Outcome Management (DOM)
Desired Outcome Management (DOM)
For the last few years I’ve been working on a system that can abstractly model long-term goals, give you ratings on how well you’re doing in pursuing those goals, and then give you practical recommendations for improvement.
I’m tentatively calling the system: Desired Outcome Management (DOM).
I’ve designed several systems that are essentially instantiations of this system, but that work at a lower level. This includes KARMA, which I have integrated into my work and captured here on my ideas page.
The idea for DOM is that it’s the highest-level abstraction of these concepts. It’s the version of the system that can be applied to anything.
Here’s how the system works:
Define your goals. This could be for a business, a family, a department, a team, or an individual. Examples are things like: graduate from a top-10 university, make 100K/year, attain 150K in passive income, have a happy and fulfilled family, etc.
Define your model. A model in this case is a method or approach for attaining a goal or set of goals. For example, if you want live a fulfilled life, there might be a Tony Robbins model, or a Dr. Phil model, or a model you make for yourself. It’ll have statements in it like, “You need to be healthy to be happy. You need to exercise. You need to eat plenty of raw foods. Etc.
Capture data. From there, you need to capture data about your entity’s behavior, from the real world, and get it into the system. So if you have a model that talks about diet, you need inputs regarding what you eat. If your model cares about grades in school, you need to get those grades into the system. Etc.
Provide Ratings. Next your system needs to provide clear ratings on how you’re doing in the various areas you’ve chosen to monitor. I prefer A through F, but you can use anything you want as long as it’s both clear and simple. Ratings will also include a composite, overall score for your progress vs. your goals.
Provide Recommendations. Finally, the system tells you exactly what to do to improve your ratings in the various areas and overall. So if you’re tracking health, for example, and you have a C in activity because you’ve been sedentary, the system will tell you what to do to improve it. It’ll give clear and prescriptive advice, such as, “Row 500 meters, do one set of pushups, and one set of situps every morning.” If you’re working on building a great team, the advice after a bad rating might be, “Have more frequent team meetings, and focus on building trust through reduced competitive focus.”
Adjustment. The last component of the system is the means by which it can be updated. Updates to the system come in the form of modifications to the model. This can be addition, subtraction, or changes in importance for elements under consideration. For example, if you’re tracking a family’s health and happiness, a new study could come out that says shared laughter is crucial to individual happiness. This will be incorporated into the model and recommendations accordingly, based not the research. And similar adjustments will be made to the model as new information about the world is made available to us.
The purpose of such a system is to continuously monitor progress towards one’s goals, and to always know what the best possible action is to take to achieve them.
A key piece of this is the weighting of the model’s components. For example, if you have a life management system, health might be weighted above popularity. Or finances might be weighted above vacations.
The key point is not which is weighted above the other (that’s up to the subject matter experts who build the model to decide), but rather that not all things are equally important in the pursuit of one’s goals. This is critical because it affects how recommendations are prioritized.
Recommendations are given as a single list of what should be done next—in order to further the goals of the entity in question. The first recommendation is always the most important, and the second is always less important than the first, but more important than the third.
Unsupervised Learning — Security, Tech, and AI in 10 minutes…
Get a weekly breakdown of what's happening in security and tech—and why it matters.
Any slight adjustment to the model will change the prioritization of the recommendations because certain elements are added, subtracted, or adjusted in importance.
When the world (or our understanding of the world) changes, the model is updated to match that difference in understanding, and with it the ratings of our progress as well as the recommendations for how best to proceed.
One key attribute that I’ve built into my implementations of the system thus far is the concept of bottom-up vs. top-down. What I mean by this is that there are a million different ways to measure, for example, health. And rather than try to measure 10,000 of them, this approach will look at the five most important things. And the same goes for team health, or financial readiness, or information security.
The idea is to start not with a list of 1,000 ideals, but rather with five must-haves. What are the things that, if you’re not doing them, will definitely result in getting a poor grade in that area? That’s where to start with any system of this kind: elimination of the things that reduce your scores the most.
Once you start scoring well there you can then add additional, higher-maturity items to the model.
Summary
DOM is a continuously updated and prescriptive system for monitoring and displaying progress towards a long-term goal, whatever those goals may be.