How to Build a Habit of Asking What Changed Since Last Time

The military concept of "ground truth" describes the actual state of a battlefield as opposed to the state represented in maps, reports, and commanders' mental models. The gap between ground truth and the model held by decision-makers has been a decisive factor in virtually every major military defeat. Plans designed against an outdated model fail not because the planners were incompetent but because they were operating with stale data that no one had updated.

This gap is not a military problem. It is a fundamental property of complex systems operating over time: models of such systems are always historical, always approximations, and always subject to drift from reality. The faster the system changes, the faster the model goes stale. Most people and organizations do not have a systematic practice for detecting and correcting this drift. They assume that the model is current until an obvious discontinuity — a crisis, a surprise failure, a shock — forces an update.

The question "what has changed since last time?" is a practice of voluntary update before the discontinuity arrives. It is cheap to ask, potentially very expensive not to.

The cognitive science underlying the problem is worth understanding. Human memory is reconstructive, not photographic. When you recall a past situation, you do not retrieve a stored recording — you reconstruct the memory using current knowledge, current expectations, and current emotional state. This means that cached models of situations are continuously being slightly revised by your current state, often in the direction of consistency with what you currently believe. The past is being quietly rewritten to match the present, which means you may not notice how much the present has actually changed, because your memory of the past has also shifted.

This is why simple comparison of "now versus then" often fails to surface real change. The "then" that you remember has been edited by the intervening time. Written records, dated and created at the time, are more reliable than memory for detecting change — which is one argument for the decision journal and the practice of writing down current assessments of ongoing situations before revisiting them.

The structural practice for building this habit has several components:

Identify your recurring situations. Most of life is recurring. Make a list of the situations you return to regularly: weekly team meetings, monthly financial reviews, ongoing projects, recurring conversations with specific people, established routines in health, creative work, and relationships. These are the domains where cached models accumulate the most drift, because they are revisited with the assumption of continuity.

Install the question as a pre-ritual. A pre-ritual is a brief, fixed action that precedes a recurring situation. Athletes have pre-performance routines; writers have writing rituals. The "what changed" question should become the pre-ritual for each recurring situation on your list. The question is asked and answered — briefly, even in a few seconds — before the situation begins. This can be done mentally, in a quick note, or verbally with a collaborator.

Build a baseline at each encounter. When you revisit a situation, your first task should be establishing what the current baseline actually is, not assuming it matches the last baseline. In practice: what are the numbers now? What do the relevant people believe now? What constraints are currently active? What is the current state of any ongoing processes? Baselines are established through inquiry and observation, not through memory.

Distinguish types of change. Not all change is equally significant. A useful taxonomy: (1) changes that require immediate action, (2) changes that require updating the model without changing action, (3) changes that are noise and require neither. Part of building the habit is developing judgment about which type a given change represents. This judgment itself improves with practice — you will develop pattern recognition for which kinds of change in a given domain tend to matter.

Track your update rate. The meta-question is: how often are you finding significant changes when you ask? If you are always finding changes, your model is drifting very fast and you are updating on a good schedule. If you are rarely finding changes, either the situation is genuinely stable or you are not asking at a fine enough resolution to surface change that is occurring. Calibrating the frequency and granularity of updates to the rate of change in the situation is itself a skill.

The organizational application of this practice is instructive for the personal application. The best intelligence analysts, doctors, and investors share a practice of explicitly separating "what I knew last time I analyzed this" from "what I know now," and actively looking for the delta. Analysts call this "updating priors." Doctors call it "following the clinical picture." Investors call it "checking the thesis." The names vary; the practice is the same.

What distinguishes the most capable practitioners in each field is not the quality of their initial analysis — though that matters — but the regularity and rigor with which they update. In rapidly changing domains, a mediocre initial analysis that is updated frequently will outperform a brilliant initial analysis that is held static.

The personal application of this is most visible in relationships. Relationships are perhaps the domain where people are most likely to operate on severely outdated models of other people. We build a model of who someone is — their needs, preferences, constraints, values — and we update it slowly and selectively, mostly on dramatic information. Meanwhile, the person continues to change: their circumstances shift, their thinking evolves, their needs transform. A friend you modeled accurately three years ago may be a significantly different person today. Without regularly asking "what has changed for you?" — not as a social pleasantry but as a genuine inquiry — you are relating to a historical version of the person, not the actual one.

The same applies to your model of yourself. One of the least examined caches is the model you carry of your own capabilities, preferences, and limitations. These were built from past experience and are continuously applied to new situations, often without recognition that you may have changed. Asking "what has changed about me since I last thought carefully about this?" — your energy, your skills, your values, your circumstances — is as important as asking it about external situations.

The habit is, at its core, a posture toward time: treating the present moment as genuinely new rather than as a continuation of the past. This is not a philosophical nicety. It is a practical orientation that keeps models close to reality, keeps strategies calibrated to current conditions, and keeps relationships grounded in who people actually are rather than who they were. It is the operational habit that makes Law 5 run continuously rather than only in moments of obvious crisis.