Guardrails, not goals

Part two of two — the build: replacing the system that enabled the drinking with systems that guardrail it.

In part one, a man audited the conditions of his life and found, on the back of a grocery list, an enabling system: a near-daily trip assembled out of an empty evening, a peak in hunger, near-zero friction, and a cost that hid in the noise of the food bill. Seeing it was the work of the audit. The obvious next step is to resolve to defeat it — to set a goal, mean it, and try. That step is the one almost everyone takes, and it is the wrong one. Here is what it looks like when it fails.

After a while my sleep was so dysfunctional that I would be wide awake for about 5 hours every night. This was after some 2 hours of dreamless sleep, and before the final 3 hours of the night, which consisted of me rolling over every 15 minutes. REM sleep was missing. I was functional (for some definition of functional) but constantly fatigued, napping often and almost always of a snide and snarky attitude.

I didn’t want to pander to myself, more to the truth was that I wanted to suffer through the episodes (perhaps even punish myself). So I would just sit in the living room until the first wave of drowsiness occurred, then I would try to fall back.

My will to change was, understandably, high during these 5 hour sessions. I would often write a note to my future self, imploring, begging, and eventually describing the hell I was going through. You would think this would affect me. It did, but not in the way you might expect. In the morning of the next day I was receptive to the note, but my will went downhill steadily and predictably, the low coinciding with the time of day when I usually sourced alcohol.

The passage above is the goal framework dying on schedule, and the schedule is the tell.

Everything the man does is what the literature of self-improvement would instruct him to do. He sets a clear intention. He makes it specific — he describes the hell in detail. He delivers it at the right time, in the middle of the night, when the resolve is genuine and total, and he leaves it for the morning, when there will be time to think about it and do something about it. The note is sincere. It is received sincerely. And then, by his own account, his will goes downhill steadily and predictably, and the low point arrives at exactly the hour he usually sources the alcohol.

That coincidence is the whole problem with goals in this domain, stated more precisely than most people ever manage to state it. The failure is not that the goal was misunderstood, or not meant, or insufficiently felt. The goal was understood, meant, and deeply felt — and the behavior happened anyway, because the goal was set in one place and the decision gets made in another. A goal is something the conscious mind holds in the morning. The drinking is decided in a hundred small moments spread across the afternoon and evening, and by the time those moments arrive the morning’s resolve has thinned to nothing, right on cue, at the bottom of the curve.

This is a structural fact, not a moral one. A society that wanted safe intersections could ask every driver, every day, to remember to slow down and watch for pedestrians. The drivers would agree, sincerely, and the intersections would stay dangerous, because the moment of approaching an intersection at speed is not a moment when resolve reliably fires. So the society does not ask. It builds a stoplight. The light does the work the instruction could not, and it does it whether the driver remembers to care or not. The man writing notes to his future self is trying, night after night, to be a better driver. What he needs is a stoplight.

One morning while looking at the list of attributes my alcohol-buying event had accumulated I had a thought. It was early, and I was sipping coffee, but, the store was open. What would happen if I did my shopping right now? At 6:15 AM?

I went to the store and did my shopping as predicted. When I was in the store, I had a very clear felt experience that it was too early, even for me, to buy alcohol. I didn’t seem to have any issues with avoiding it; my ability to avoid was high — and I recognized that if my ability to avoid changes throughout the day, then it must have a high spot and a low spot. Shopping after work hit the low spot. Shopping first thing in the morning hit the high spot.

When I got back home, after putting the groceries away, I re-looked at the list of attributes. By simply changing when I shopped, I was able to materially change the default enabling system that was in place. I now had two systems: one for shopping (that entailed guardrails that kept me on track), and the existing after work shopping system; and this system had been altered — the list of attributes was smaller — which meant that the pull on me was weaker.

I wasn’t yet out of the woods, but now I had a working model, and a direction in which to tweak it.

The passage above is the man building his first stoplight, and it is worth taking apart, because the levers available to anyone doing this work are all visible inside a single change of hour.

A guardrail — a system, in our sense — is a deliberately arranged condition under which the behavior you want is the path of least resistance and the behavior you do not want carries more friction. It does not require you to win the moment of temptation, because it has been arranged before the moment arrives, and it runs in the background whether you are watching it or not. The man does not add a gram of willpower. He moves one variable — when — and lets the variable do the work that the nightly notes could not.

Watch which levers he is pulling. The first is sequencing: what happens in what order. He has discovered, by direct experiment, that his capacity to decline is not a fixed quantity but a curve with a high point and a low point across the day, and that his old routine had quietly scheduled the purchase at the low point — after work, hungry, depleted. Move the same act to six in the morning and it now falls at the high point, where, in his own words, declining costs him nothing. This is the most transferable thing in the passage: you can almost always relocate your highest-clarity hour to the place where your highest-stakes decision gets made. The second lever is the default option — what happens if you do nothing at all. He has not banned the alcohol. He has simply rearranged the day so that the unconsidered, do-nothing version of it no longer routes through the beer aisle. Most of the not-drinking in a well-built system is not the product of resistance. It is the product of a changed default.

Then notice what he says he now has: two systems. The old after-work system, weakened — its list of enabling attributes shorter, its pull correspondingly weaker — and a new morning system that holds. This is the opening of the most important principle in the whole enterprise. A single mechanism, however strong, is fragile. Walker’s first sober stretch, years before this one, ran for four and a half years on one mechanism alone, and it collapsed completely the day that mechanism was finally breached, because there was no second line behind it. A durable structure is not one strong wall. It is several walls, arranged so that the failure of any one does not bring down the rest. He has just built his second, and he has noticed, correctly, that he now has a working model rather than a resolution.

Once I realized I could break the post-work shopping enabling system into two systems: the early morning guardrail system, and the weaker post-work system, I realized that this was not a one-time event, it was a concept. So I set to work on trying to address the attributes of the post-work shopping system to dismantle it by putting guardrail systems in its place.

As I developed the system of systems, it wasn’t a binary switch that turned from off to on. In the beginning, I was still very susceptible to discovering loopholes to my system.

One day after work I noticed that I had forgotten an item I needed at the store. Flush with the confidence that the new early-morning shopping had given me I impulsively decided that I would just hit the store really quickly. It all happened so fast and without any foresight, and predictably I was in the beer aisle again.

The next morning, as disappointed as I was in myself, I had a further realization. Long ago I noticed that when I called a bike ride “bad” it led to me just feeling crummy; however if I termed the bike ride “too cold” then every changed, it spurred me to realize that I could enjoy the ride more with better attention to weather and clothing.

As such, better than terming the alcohol event a “failure”, a more constructive approach was to term it “information”. My system of systems was evolving with inputs of information, which led to better systems. Sitting there, I crafted a pre-shopping meta checklist, which included: check the frig, check all cabinets, count items so that you can stock X days worth of food, put wallet in pocket, etc. This specific meta-checklist system allowed me to address the situation to which the information pointed.

Yes I had “failed” in the classic sense. Instead of shame, however, I updated my systems. When I look back on the audit trail, what I see is a progressively mature system of systems, that at one point tipped such that enough systems were in place and I remained sober.

The passage above is the system being stress-tested by the only thing that can really test it — ordinary life — and it shows the man doing the single most important thing a person in this work can learn to do.

He forgot an item. He went back after work, into the low point of the curve, on a burst of misplaced confidence, and he ended up exactly where the old system used to deliver him. Under the older framework this is a relapse: a failure, an occasion for shame, and shame is the trapdoor through which the entire moral model comes flooding back in. The man declines the trapdoor. He calls the event information instead. This is not a comforting reframe; it is the correct engineering description. A system reveals its weak points only under load. The forgotten item was a load test, and it returned a specific, usable result — there is an unguarded path back to the low point, opened by a gap in his shopping routine. So he closes the gap. He builds a pre-shopping checklist, fitted precisely to the failure that exposed the need for it: a new guardrail, manufactured out of a lapse.

This is how a system of systems actually comes into being. Not as a switch thrown once, but as a structure that accumulates — each loophole discovered, named, and designed around, each failure converted into the next guardrail. The written list he keeps returning to over his morning coffee is doing quiet double duty in all of this: it is not only the record of the original audit, it is the reference against which he notices drift and to which he adds each new defense. The mind that built the system is also the mind that monitors it, and the page is what keeps the monitoring honest.

And then, in his last line, the thing tips. At one point it tipped such that enough systems were in place and I remained sober. There is no single heroic morning in that sentence. No white-knuckled victory. There is a structure that grew, guardrail by guardrail, until the drinking simply had no easy path left through it. That is what the work produces — not a fight bravely won each day, but a life arranged so that the fight mostly does not arise.

It is worth being clear, at the close, about what all of this is for, because it is not for the prevention of drinking. The guardrails are not a cage, and they are not a deprivation. They exist to hold the conditions steady long enough for the hedonic system to recalibrate — long enough that a person falls back in love with the unmodified life, and the substance quietly loses its pull on its own. The destination is not a man heroically resisting the beer aisle for the rest of his days. It is a man walking the length of a strip mall, passing store after store he has no reason to enter, and passing the one that used to own him with no more thought than he gives the others. No defect. No drama. Just one more thing, among the thousands, that has nothing to do with him.

— Ben Walker