Devising means of measurement will be left as an exercise to the reader, however I recommend that serious thought and effort be given to such testing.
I expect this is where much of the challenge and nuance of this technique lies. A few questions come to mind when I try this. How do you handle terminal vs instrumental value? How do you handle various kinds of uncertainty? (expected value calculations can require your whole world-model if you don't limit scope!)
Do you have any specific examples of how you perform this measurement on a real decision?
I assign weights to terminal and instrumental value differently, with instrumental value growing higher for steps that are less removed from producing terminal value and/or for steps that won't easily backslide/revert without maintenance.
As far as uncertainty goes, my general formula is to focus upon keeping plans composed of "sure bet" steps if the risk of failure is high, but I'll allow less surefire steps to be attempted if there is more wiggle room in play. This sometimes results in plans that are overly circuitous, but resistant to common points of failure. The success rate of a step is calculated from my relevant experience and practice levels, as well as awareness of any relevant environmental factors. The actual weights were developed through iteration, and are likely specific to my framework.
Here's a real example of a decision calculation, as requested:
Scenario: I'm driving home from work, and need to pick which restaurant to get dinner from.
Value Categories (a sampling):
I don't enjoy driving much, so Travel Distance is usually the highest-ranked Value Category, thoroughly eliminating food options that are too much of a deviation from my route. Next is Existing Desires, then Diminishing Returns, which let me pursue my desires and avoid getting overexposed to things. My finances are generally in a state where Price Tag doesn't make much difference on location selection, but it will play a more noticeable role when it comes time to figure out my order.
Decisions are a part of life that are frequently targeted for optimization by normal people and rationalists alike. In particular, decision paralysis often causes processing delays that can cripple attempts to streamline your life. Systematically removing decision paralysis from your life is an excellent step on the path to self-improvement, although the task can be daunting to consider.
This guide is based upon my own solution to decision paralysis and the lengths to which I pushed it, but its requirements can be rather daunting to a novice rationalist or a normal person.
-Value Category: A specific type of value which you consider worth obtaining or avoiding (negative value is still value!), isolated from other Value Categories you track as much as is reasonable
-Course of Action: An option that you have identified through analysis of the decision being made and found to be worth keeping distinct from other available Courses of Action from the decision point
-Estimate of Gained Value: An estimate of the value you expect to derive from following a Course of Action, which may be specific to a single Value Category or may be a general estimate overall
-Value Table: A table containing an Estimate of Gained Value for each Value Category for each Course of Action you are considering in a decision
You must have a healthy knowledge of the Value Categories that you expect to derive from common choices. Money is a popular variety of value to pay attention to, but this list should include anything you consider worth obtaining or avoiding (negative value is still value!). There should be a fair number of guides and articles that can help you generate this information, so I shall leave the task to them. I am largely self-taught on this subject, so it is difficult for me to provide citations.
You will want to be familiar with cementing habits into your muscle memory, be it physical and/or mental. This exercise will involve embedding a complex program into your reflexes, and if you are unable to perform such a change without losing some of the intricacies then you are unlikely to find success with this technique until that matter is rectified. I will somewhat discuss the methodology in broad strokes, but it is unlikely I will provide enough details for readers to learn the technique from scratch.
Folks with a math or computing background are likely to spot opportunities to use calculators to shortcut some of the following steps, but I must emphasize that performing the calculations in your own head is critical to the process. It is not even required to use actual numbers for any of the calculations that will be described, so long as you are able to assign relative priorities to your types of value.
These milestones are in ascending order of difficulty, and may not be reached in even intervals.
You should be able to:
-Rank your Value Categories by their relative value at the time of evaluation (a caching strategy can be found in the Extra Credit section)
-Generate a Value Table whose Value Category columns are in sorted order based upon your generated ranking
-Use a sorted Value Table to generate a calculated answer
-Update input information (estimation methods, Value Category list/ranking) if the calculated answers show consistent bias worth correcting
-Perform the calculation whenever a decision appears so that it can become reflexive
You should be able to observe and measure the following:
-Your calculated answers generally align with the answer you would have chosen using more traditional decision-making methods
-Your calculated answers should spring to mind nearly instantaneously once you have sufficient information
-You should immediately notice if you are missing information required to make the decision automatically
-You should be able to trace your decision-making calculations to generate explanations for people who question your calculated answer (they don't necessarily need to hear the true explanation, but you need to be able to retrieve it before you develop whatever you're going to tell them)
Devising means of measurement will be left as an exercise to the reader, however I recommend that serious thought and effort be given to such testing.
As with most efforts to program your reflexes, you will spend a significant length of time repeating and iterating upon a basic action. The action in this case is to calculate Estimates of Gained Value by Value Category for each available Course of Action that you are considering. This will allow you to produce a quantified ranking of the courses of action, at which point it becomes difficult to spend further time on this part of the decision-making process.
When practicing this process, it is likely that you will find yourself wanting to take notes or take other steps to assist yourself in performing the calculations. It is perfectly acceptable and even recommended to start off in such a way, so long as the additional measures do not become a hindrance. That point of diminishing returns will vary from person to person, so I shall leave its identification as an exercise to the individual.
Bringing to mind your list of value categories, generate a ranking of the different categories by whatever method you consider appropriate. It is acceptable to have multiple types of value considered to have equal overall value, so do not expend much effort trying to resolve any ties you may encounter. It is important to attempt this ranking process every time that you perform the process, so that your resulting reflexive actions will be self-correcting.
Once you have generated the ranking of value categories, step through the list of value categories and estimate the quantity of that value that you expect to gain from the Course of Action you are currently considering. If you are taking notes on paper or a computer, this is best stored in a table format as a single row, with a column for each Value Category.
Repeat this process for each Course of Action, making sure to write down the results if you're actually writing things down.
Using your Sorted Value Table, start sorting the Course of Action rows by the overall value you have estimated them to provide. At first glance, this sounds simple, as you can simply arrange the rows based upon the highest-ranked value quantity that they provide. However, to properly perform this estimation, you will need to estimate a "rate of exchange" between Value Categories. If you feel that no amount of value from one Value Category can match the value of a single unit of value in a higher Value Category, simply treat that rate of exchange as ridiculously high, so high that it cannot be achieved. These rates of exchange should be calculated anew each time you face a new decision so that they can reflect your current circumstances.
The resulting ranking should allow you to make your decision with minimal delay. If there is insufficient need to doubt your estimations, it is typical to simply choose the Course of Action that was estimated to be most valuable. However, it is often beneficial to perform a "sniff test" of the results for a while until you have properly calibrated your estimation skills and Value Categories if they are not already sufficiently polished.
The most important step in the development of any tool or technique is to measure its effectiveness. In this case, there are typically three main points of failure that may emerge:
-Estimates generated were not very accurate, in which case you should study the gap between estimate and reality briefly and then continue practicing estimation
-Some Value Category was not considered that would have drastically changed the results, in which case an effort should be made to study the Value Category (if necessary) and to include it in future iterations of the calculation
-The Value Category ranking was not accurate or failed to be updated from a previous iteration, in which case an effort should be made to re-generate the ranking when performing the next iteration of the calculation.
Many of the more prominent issues with embedding programs in your reflexes arise from a lack of inspection and/or maintenance. Fortunately, I have provided a Definition of Done that serves as a template for any regression testing you may develop, and I recommend performing such a regression test at least once a month.
Generally, there are two types of issues that will appear in the long run:
-The Definition of Done was not actually met or correctly evaluated originally, and the technique is therefore flawed
-Some part of the calculation is not operating correctly, and the results are coming out obviously skewed
In the first case, it will likely be necessary to start at the beginning of this process and perform the embedding process again in full to overwrite the existing muscle memory. It may be tempting to try and "patch" the embedded program, but such changes are often quite difficult to apply safely even if the problem seems to be a simple one. The attempt to patch will typically result in partial corruption to other parts of the process, at which point you'll have to re-embed anyways.
The second case implies some sort of calculation error and should be investigated via a thorough stacktrace as quickly as possible. Such errors can often propagate quickly and can also prove difficult to address once you have identified them, so be ready to encounter serious obstacles. In computing terms, this could be an indicator of issues with the host OS or the host hardware if there does not turn out to be a logic error in your iteration of the process.
For the ambitious, there is a way to use caching that can allow you to reduce the processing burden of ranking your Value Categories. The heart of the technique is to identify when you receive information that would affect your Value Category ranking, and to estimate the effects at that moment. This update can be as simple as moving a single Value Category up or down the ranking, or a complete recalculation if the new information changes things to that degree.
Take as an example a Value Category that almost never sees any positive value from Courses of Action you encounter, and you have placed it very high on the ranking as a result. If you found yourself in a new environment where it became nearly trivial to achieve positive value from that Value Category, you would likely find yourself adjusting its position on the ranking in the near future as the relative scarcity fades.
I perform decision calculations more often than I receive information that affects my Value Category ranking, so this optimization is worthwhile for me to use. If you find yourself experiencing a different ratio between the two frequencies, then you should perform your own value estimations before attempting to integrate.
This is my first venture into writing for a rationalist audience, or really any audience at large. I recognize that this guide is likely to be unhelpful or frustrating for those who have not already spent a large amount of time reprogramming their reflexes and clarifying what value categories they really pay attention to, but I did not find any guides that were specialized in this direction in my first several days of browsing the forum.