XVI. Level Up

The fundamental principle of planning and execution for complexity and uncertainty.
Dr. Albert Künstler

raditional project management has a simple yet long established view on task completion: initiated tasks are either “in progress” or “done”. If a task is large enough, then it should be broken down into small parts. It stays “in progress” until all its parts are “done”. Such a tradition has its roots in and works quite well for ordinary problems.

However, as discussed earlier, there are a large number of strong problems. If we try to apply the old tradition to strong problems, they will remain “in progress” for weeks and months, and will never give us the joy of seeing them “done”.

Disaggregation will not cure us of this trouble, because you might be terrified to find that a strong problem breaks down into several smaller, but also strong, sub-problems. And these have no predictable completion time, but rather a risk of being trapped “in progress” forever.

Traditional “In-Progress / Done” way of thinking leads to disaster

In this chapter, we will discuss a fundamentally different planning and execution approach. It must be very well understood and accepted by both the heroes and the quest givers. If at least one person acts in the old way, the whole endeavor is ruined.

Principle: Level Up — instead of planning a binary (undone → done) completion of work items and system components, plan gradual completion at increasing levels of fidelity.

Key concepts: Level up, Zero, Placeholder, Quick & Dirty, Good Enough, State of the Art, Castle in the Clouds, Fidelity

Observed in: Planning, Estimating cost, Prioritizing work

So, we are at the commencement of the quest when the sponsors have explained the challenge and want to hear the action plan and know the cost. The old approach advises us to disaggregate the problem, prepare a scope of work, evaluate it, add some risk, and come to an agreement.

This is a valid approach if there are only ordinary problems in the scope.

However if one, two, or three strong problems fall into the scope, which, as we know, cannot be “100% completed”, then already from the very beginning we are in trouble. The captain feels uncertainty, increases the price in order to cover unforeseen expenses and tries to get rid of this “risky” problem, putting it “out of scope”. The sponsor, in contrast, always insists on lowering the cost and as the very nature of the anticipated quest requires this strong problem to be solved, insists on keeping it “in scope”.

In the end, the sponsor wins and the parties sign a contract with a sleeping dragon inside: the strong problem aislisted to be solved and required for acceptance. Many months later, well behind schedule, the parties meet again, look at the contract, and are forced to admit that the project has failed because this essential item has not been fully completed. The captain can argue: “we did a lot”, the sponsor will respond: “not as much as we expected”. But legally and technically, this item of the contract was not “completed” and not “delivered”.

The sponsor “wins” the contention, but at what cost? Nobody is happy, both sides have lost many months and substantial amounts of money, the crew is fired, their findings are nullified, the problem still exists, and the sponsors will need to find another crew and start the cycle over.

There is a better way.

The Level Up way of thinking, planning, and execution

The creation of something new begins and goes through gradual stages from simple forms to more complex and perfect ones.

For sophisticated tasks, strong problems, and complex system components, consider a “Level Up” approach. Instead of spending an eternity on finding the “best solution in theory”, start with the simplest, even imperfect solution. Use it. Improve it. Then collect more data, do some exploration, and upgrade it again. Apply it to a broader range of situations, spend more time on polishing and improving it again, and gradually get a reliable robust solution, which is in fact the “best solution in practice”.

The advantage of this approach is clear. You don’t need to wait until the task is “100% completed” — that will never happen. Instead, you can enjoy the benefits of temporary solutions almost immediately.

Although the full spectrum is uncountable, we will distinguish five levels of creation:

Zero — nothing has been created yet.

Placeholder — there is an idea of shape and place of the solution in the system, but for now it is only an unusable stub.

Quick & Dirty — the idea is starting to emerge, although it is still very imperfect, it can already be utilized, if handled with care.

Good Enough — the solution grows mature, reliable, very practical, not inferior to the counterparts.

State of the Art — the solution is unique, excels in all things, and leaves the other existing solutions behind.

5 levels of solution.

These levels are distinguished for a good reason. They not only speak of quantitative progress, but are also responsible for important qualitative stages of creation.

Zero: high uncertainty, not enough understanding of importance or structure

Placeholder: the problem is understood and prioritized, no solution yet, but there are some initial vague ideas and designs. Placeholders can not be used to solve a real problem, but they can help explain the idea of how the real problem could have been solved. Examples of placeholders: sketches, mockups, back-of-the-napkin calculations, early assumptions, dummies, stubs, paper prototypes, first research results taken from the internet.

Quick & Dirty: the first actionable version of the solution. It has quite a narrow range of applicability. It has many flaws and may easily disenchant the demanding consumer. But it works and solves at least some most crucial manifestations of the target problem! Examples of quick-and-dirty solutions: a working prototype, MVP, assumption validated by 4–10 experimental trials (below statistical significance), temporarily manual procedure of performing an operation, concierge service, boss’s opinion taken for an assumption.

Good Enough: stable and reliable version of the solution, as good as the other similar solutions in the market. It is no better, but it is no worse either. It may not solve the problem completely but it provides a robust solution for the most common situations, without guarantee to handle rare and exotic events. Examples of good enough solutions: 80% of all existing systems, de-facto standard solutions, most common products around us in our life, hypotheses validated by 200–400 experimental trials (achieving statistical significance), well-tested and documented automation procedures, carefully evaluated alternatives, deep literature search and analysis, boss’s decision supported by extensive data collection and expert knowledge.

State of the Art: an unusual and extremely efficient solution that outperforms all existing counterparts. The disruptive technology at its best application with a lot of practical value. Unique and exorbitant level never achieved before for this type of problem. Examples of state of the art solutions: Top-5% of all markets, top grossing products, booming tech, bestsellers, notable discoveries and inventions, major cost reduction or profit driving improvements in products or services, unexpected outstanding results of the series of experiments, a game-changing visionary boss’s decision supported by data, luck, and vision.

As you can see from the list above, the same task can be solved at a completely different level of depth and effect. In the binary world of “done/unfinished” — when we don’t specify the level of implementation, we have a conflict. Since the sponsor may expect “state-of-the-art” at the price of “quick-and-dirty” while the crew aims for “good-enough” and expects enough budget for this level.

But how fortunate it is if the sponsor and the crew share the same understanding of the complexity of the challenge. With the “level-up” mindset, they get rid of a binary understanding of task completion and move on to more versatile planning and prioritization. Where necessary — using placeholders, where a reliable process is a must — implementing it out of “good enough” components, where experimentation is required — constructing “quick and dirty” prototypes, and thus more efficiently planning and using the resources to succeed on the quest.

Linear and Fractal planning

Understanding the “5 shades of done” doesn’t make us masters at planning strong problems. The mistake we can make is simply renaming the old words: instead of “Step 1:done, Step 2: done, Step three: In-progress”, use some new fancy names: “Placeholder: done, Quick-and-Dirty: done, Good-Enough: In-progress”.

This linear step1-step2-step3 approach works well for delivery projects and tasks. It’s totally fine to keep using the old “step-by-step” and “in-progress/done” way of planning and managing standardized, predictable, and low-uncertainty delivery activities.

It’s ok to use linear planning for predictable delivery activities.

However, when it comes to pathfinding for uncertain, research, and strong problems, a fractal “Level Up” approach is just what is needed.

There are three important principles.

1. Castle in the Clouds: Set up soft commitments.

One cannot demand a hard commitment to 100%-complete or a State of the Art solution to a strong problem. A State of the Art solution is a breakthrough invention, the result of luck and hundreds or thousands of experiments, and a worthy long-term objective, but it may be only required with very generous funding and unrestricted time resources. That’s why the sponsors should designate their long-term objectives as a Castle in the Clouds: something desirable, but that needs to be discovered throughout the quest. It might be a mirage, but it may be the real thing. Adventurers will have this priority before their eyes and do their best to find the path.

2. Pathfinding: Explore several lines at a time.

In the labyrinth, direct and linear paths do not work. If you want to reach a distant Castle in the Clouds, it is not enough just to move in its direction, because at some point you will hit a dead end. Instead, you need to choose 3 or 4 smaller objectives of a lower level and try to direct your crew to explore them. For example, develop not one but several Quick & Dirty prototypes that implement alternative ways to solve the target problem. Some of them will be movement towards a dead end, but others will have potential that can be developed by refining them to a Good Enough level. These components, as well as the knowledge gained while working, can be combined in order to find a further path to the cloud castle.

3. Level Up rules: Upgrade what makes the most impact, but also keep experimenting with new ideas.

Components that show the most impact deserve a Level Up, and those that don’t are sorted out. However, do not strive for excessive efficiency. Discarding a seemingly unsuccessful idea, you can trample the sprout of a future non-standard solution. Therefore, develop and upgrade ideas that grow and shine, but always keep incubating a set of new ones even stranger and ridiculous. Don’t kill them while they’re still on the vine.

Level Up technique in practice

Let’s take a look at how a case could evolve with the Level Up technique. In Chapter XII, we used the Decision-to-be-Made tool and discussed the Decision Logic of a hypothetical example. As we learned then, businesses make decisions by choosing from three options. Option C is chosen the least often, only 17% of the time, but it is associated with the highest cost of errors for the business.

Decision logic example from Chapter XII

To begin with, let’s define the long-term objective in the reduction of mistakes with option C — this is our Castle in the Clouds. How do we do that? In order to reduce these mistakes, we need to look at the previous steps in the process that lead to this decision. Can we improve them somehow? Can we use more data, prepare it better, or carry out additional checks? These will be partial solutions that we need to design and implement.

In the first month, we will study these steps and outline Placeholders where our improvements will appear.

In the second month, we are trying to develop Quick & Dirty prototypes of solutions for the two steps of the process.

In the third month, we upgrade one of the prototypes to a Good Enough level and add new Quick & Dirty prototypes. At this moment, we are already starting to get the first insignificant business results and the number of errors when choosing option C is slightly reduced. While not as much as we wanted yet, we can already admire the first observable results.

In the fourth month, we improve the level of the developed solutions at several process steps at once, this gives a Good Enough reduction in mistakes on the C option and adds confidence to our quest givers in the successful course of our quest.

Some may argue that gradual improvement described in this chapter is no different from “in-progress” works, inasmuch as “Good Enough” is the minimum acceptable level of quality that deserved to be put in front of clients. So, everything before “Good Enough” is “In-Progress”, and everything after — “Done”.

This is a good sentiment, but surprisingly, “Quick & Dirty” solutions and even “Placeholders” can also be introduced to the clients.

How is that possible? The customers will not tolerate deplorable quality!

But in fact, “Quick & Dirty” solution does not necessarily mean poor quality.It only means our solution is very limited and may not serve us well in any situation.

Quality and Fidelity

Quality is important.

As we are discussing for-profit research quests, we always need to think about customers and make sure our solutions do no harm, and bring no frustration to the customers.

Let us introduce the idea of Quality vs Fidelity to see the full spectrum of possible solutions.

We will use the term Quality in a common sense. A low-quality product is “bad” — it’s ugly, it breaks all the time, and it requires constant tinkering. High-quality products are good looking, they have no “bugs”, they deserve to be paid for. But a high-quality product does not always mean a useful product.

We will use the term Fidelity to describe how capable the product is in solving the problem. A low-fidelity product does not solve the problem, or solves the problem partially and in a very narrow range of situations. A high-fidelity product offers a universal solution that is applicable in various situations and can address multiple manifestations of the problem.

Improving both Quality and Fidelity requires a lot of effort, but you may have them in different proportions.

A Low Fidelity / Low Quality product is an “Ugly baby” — it is primitive, dysfunctional, and unreliable.

A Low Fidelity / High Quality product is a “Dexterous apprentice” — it is not very useful, but it is elegant, simple, and dutiful.

A High Fidelity / Low Quality product is an “Unrecognized genius” — it is remarkably intelligent and capable of solving the problem, but it’s scrappy and unkempt, not designed for humans, and only its creator knows how to make it work.

A High Fidelity / High Quality product is a “Mighty hero” — this is exactly what everybody wants, as it is a reliable and efficient product that has been designed and tested for problems and users of all sorts and kinds.

Knowing these differences, you may choose your leveling strategy. A common mistake is to develop both Quality and Fidelity simultaneously. This “middle way” strategy is bad. For many months your product will stay in the “Ugly baby” quadrant and annoy the quest giver.

The “middle way” and the “nerdy way” leveling strategies. Neither of these are recommended.

Another mistaken leveling strategy is the “nerdy way” through the “Unrecognized genius” quadrant. This strategy is super-popular among scientists who came into commercial research from academia. Once they get their hands on a problem, they work on it day and night, striving to find a universal solution and forget about customers completely. By the end of the day they have a great solution that nobody understands and nobody can use. Though this kind of solution may be well appreciated in an academic community and a non-profit world, the quest givers will be embarrassed.

The most recommended leveling strategy is the “elegant simplification way”: prioritizing Quality over Fidelity. Yes, you limit the ability of the product to solve the problem, but you make it user-friendly and even a Quick & Dirty version of the product is no more an “Ugly baby”. With this leveling strategy, a Quick & Dirty solution is just a radical simplification of the problem — it looks good, it works, it has no glitches, but it’s made very simple.

Elegant simplification in action.

The quest givers like this way. They receive the first results early, and even though they see the limitations of the Quick & Dirty solution, they are amazed with its simplicity and elegance, so they understand the future versions will follow the same standards of quality, and the customers will be taken care of.

For the crew this way is also rewarding. Increasing Fidelity is hard. It is the journey through the land of uncertainty and requires an unpredictable amount of resources. Increasing Quality is much easier. Yes, you’ll need to think about it, educate your academia researchers about being customer-centric, probably you’ll need a quality assurance or user experience person on your team, but this is much more predictable and less expensive. After all, it will keep your sponsors inspired.