Another reflection in my optimization quest: Speed and Precision
When is it better to work as fast and hard as possible, and when instead to take a step back and look at the big picture? 🤔
No amount of hard work will take you to your destination, if you don't know where it is.
No amount of thinking will move you forward, if you don't actually move.
A few notes on moving between these two modes 👇
At the extremes: overthinking and burnout
Let's look at the limits before finding a balance.
When you stress about precision, it's called overthinking.
Precision requires standing still, trading speed (
=0) for a promise of a future higher acceleration 🙌
In practice, planning doesn't move the needle forward. Practice does. Too much thinking, and you fall into "procrastination" land.
On the other extreme, speed only leads to more learning, but with more energy expenditure and with the risk of reaching the wrong place 🤔
Speed is "safer": you are guaranteed a payoff, eventually. But, if planning is 0, you don't know what payoff and when. And if the payoff doesn't show up, burnout is a risk (when sustained effort fails to produce meaningful progress).
Rule of thumb: when in doubt, choose speed.
Volume negates luck.
The case for precision
Precision means finding the highest-leverage problem to work on.
Fixing the right problem leads to more progress than speeding ahead on meaningless tasks 🫡
Precision comes before the actual "fix". It's about finding the problem, not working on it. In other words, standing still, to then move faster later.
Standing still is expensive,
speed = 0, and the result is not guaranteed 🤯
The solution is trading "perfection" (finding the single most important problem) for execution (finding one important problem).
The goal is not “find the perfect problem”, but reduce uncertainty until one problem dominates by evidence
Rules:
- Time-boxing: fixed time for reflection, after it either (1) commit or (2) run an (any) experiment
- Start from friction: reflection is valid when something is off, so that you know where to search
- Hypotheses: come up with a written bottleneck, how to exactly test the hypotheses, and a success criteria
- Cheap falsification over deep confirmation: the goal is to be wrong quickly
Remember: precision must increase future speed, otherwise you just moved work around 🫵
The case for speed
Speed is ideal when learning or delivery dominates optimization.
Practice is the condition for learning, increase speed when learning is the priority 👀
Speed aims to verify/falsify an hypothesis, either outcome leading to a new insight. As such, speed must be scoped.
Define what you are not doing. Define the smallest version that creates signal. Then full speed ahead ⚡️
Another key: speed is not only hard work. Speed must produce new knowledge. Forward movement without progress is fake speed.
When to choose speed:
- High uncertainty
- Low cost of being wrong
- You are blocked by ambiguity, not complexity
Precision and speed in loop
Here is the dance of precision and speed in practice:
- Choose something to work on (anything at the beginning)
- Work on that as fast as possible until no new insights
- Stop, reflect on what's not working, define an hypothesis
- Work fast to verify/falsify the hypothesis
- Stop, document what you learned, move to the next hypothesis
Over time, insights accumulate, leverage will increase. Also, reflection will need to become more precise to spot issues in the details.
My initial intention was to title this "Precision over speed". But that's not correct, it's a bias based on where I find myself in the cycle right now.
The right answer is always nuanced, don't fall trap of cheap/generic advices.
See you next 👋