The Tuesday-Thursday problem.
On the schooling decision Alex and I could not settle at the dinner table, the answer was obviously yes on Tuesday and obviously no on Thursday. The difference between the two evenings was not new information; the difference was us. This is the long version of how that observation, sat with for long enough, became Counsel.day.
A school open evening, and then a hill with a dog.
The question was whether to put our son into the city school the following year. The school was good. The trip was forty minutes, each way, on a road that was not always kind. We had been circling the question for the better part of two months and could not put it down. We sat with it through summer, we sat with it through autumn, we sat with it over more dinners than I want to count, and we kept arriving at different answers depending on the day.
On the Tuesday I am thinking of, we had gone to the school's open evening. The classrooms were small and the children were attentive and the head spoke about reading and the library had a window that looked out over a clay-court tennis hall that you could imagine an eight-year-old being lit up by. Alex and I came home and sat at the kitchen table and said, more or less in unison, of course we are doing this. The answer was so obvious it felt embarrassing that we had been circling it. We poured a second glass.
On the Thursday, two days later, our son spent three hours after school sliding down a hill in the Canterbury countryside on the back of the family dog · which I realise is not a sentence that makes sense if you have not lived on a small lifestyle block, but trust me, it was a magnificent thing to watch. He came in covered in mud and barefoot and ten years old in every direction, and Alex and I sat down to dinner and said, more or less in unison, of course we are not doing this. Forty minutes on the road each way, every morning, for a child whose life was this. The answer was obvious.
Both signals were real. Neither was the answer.
The thing that has stayed with me about that week is that nothing had changed between Tuesday and Thursday in any factual sense. The school was as good on Thursday as it had been on Tuesday. The hill and the dog and the muddy boy were as wonderful on Tuesday as they would be on Thursday. We had not learned anything new about the road, or the bus timetable, or the cost. The information was constant. We were not.
On Tuesday I was the man who had just watched my son being interested in things in a classroom, and I was full of the possibility of him being formed by good teachers and a library and a tennis court. On Thursday I was the man who had just watched my son being a child in his own field, and I was full of the cost of putting him on a road that would take him out of it. Both men were me. Both were honest. They voted in opposite directions on the same question, two days apart, on the same evidence.
And then, the question being unresolved, we would have a third night at the dinner table to settle it. Whoever spoke first would set the anchor for the conversation. Whoever was tireder would yield. We would arrive at an answer that was a synthesis of two people performing reasonableness, rather than a synthesis of what we each actually thought. And the question would come back the next month, because we had not actually decided it; we had only stopped talking about it.
The information was constant. We were not.
The literature has names for this.
I am not a cognitive scientist; I am a data professional who spent twenty years watching dashboards lie to people who were sure they were honest readers of them. But I went and read, because the Tuesday-Thursday pattern was familiar in the way that all good observations are familiar before they are useful: I had seen it in every team I had ever run, and I had assumed it was a failure of the people involved rather than a feature of the brain.
The names exist. Affective forecasting tells you that the version of you who is making the decision is not the version who will live with it. Mood-congruent decision making says that your read of the same facts shifts with your emotional state. Hot and cold cognition (Loewenstein, others) describes the gap between the system that decides in a moment of arousal and the system that decides at a remove. System 1 and System 2 (Kahneman) is the most famous form of the same intuition. Couples therapists talk about pseudo-agreement: two people who have stopped arguing because they are tired, not because they have agreed.
None of these are pathologies. They are how a working brain handles a hard question. The mistake is not the variability; the mistake is treating any single evening's reading as the verdict, when the variability is the data.
If the data is the arc, build the arc.
The product implication landed for me on a run, a few weeks after the schooling thing eventually got decided (yes, in the end, for reasons that had less to do with the school than with the road). If a single evening's answer is unreliable, and the unreliability is itself informative, then the right move is not to find the magic single evening. The right move is to capture every evening, on the same question, across a period long enough that the moods cancel out, and read the arc.
Two people, voting privately, every evening, for a duration set by the weight of the question. Neither sees the other's vote until the period ends. The verdict on the final day is not a single answer; it is the trajectory, the agreement rate, the themes that surfaced in the notes, and a synthesis of what we were actually arguing about underneath the surface question. The synthesis is the load-bearing piece. Most couples are not arguing about whether to do the thing. They are arguing about who would carry the cost; what each of them would have to give up; whose version of the future the other one is asked to live in. The data will show that, if you collect enough of it.
The product is built on that observation, with the privacy enforced underneath. Neither partner can read the other's drift mid-period, because if either could, the anchoring would reappear inside the product. The sealed-vote architecture · row-level security on the votes table, per-decision keys, no admin override · exists for exactly the reason the dinner table fails: the moment one of you knows where the other is leaning, the other one stops being honest with themselves. The whole machine depends on the seal.
A small caveat about what this is.
Counsel.day is not therapy. I am not a therapist, I have no clinical training, and the product is not validated, built, tested, or endorsed by clinicians. What it is is a data structure built around an observation that anyone who has tried to settle a hard joint question across a dinner table will recognise. The structure is mine. The deciding stays with you.
If, having walked through a season with the product, the verdict reveals that the decision sits on something larger than the decision · something about who you each are, or what you each want, or what you each cannot bear · then the conversation that follows is one for a therapist, not a verdict page. Many of our users will be in both places at once, with our verdict in one hand and a clinician in the other room. We have no objection to that; we built the product to live alongside that kind of work, not to replace it.
A decision worth losing sleep over deserves the whole arc, not the last evening you were tired enough to settle it.