that is a slightly absurd sentence and i stand by it. the honest reason was not finance. it was that the question would not leave me alone, and the only way i know to really settle one like that is to build the smallest world where it is the only thing that can vary, and then watch.
so i wrote a market in go. a real one, in miniature. an order book, the matching rules that decide who trades with whom and in what order, the whole price-time priority that actual exchanges run on, where if two people want the same thing the better price wins and ties go to whoever was there first. then i put two traders inside it running the exact same strategy, identical in every way but one. one of them sees the market a few milliseconds sooner.
everything else i held still on purpose, because that is the whole trick of building the world yourself. in the real one you can never isolate anything, a thousand things move at once and you are left arguing about which mattered. in a built one you freeze all of it and turn a single dial. so the only difference between my two traders was time, and anything that showed up in the results had exactly one thing to blame.
then i ran it through different weather. a calm book with plenty of orders resting on both sides, a thin one where liquidity is scarce and every move costs you, a spiky one where the market lurches. same two traders, same single difference, three kinds of day, scored on the things a trader actually cares about, how often they got filled, how far from the quoted price they ended up, how often they were the one caught on the wrong side of someone better informed, where they sat in the queue.
the part that stayed with me was how quiet the advantage is. it does not announce itself. the faster trader is not winning huge dramatic trades. it is sitting a little further forward in the queue, getting filled a little more often, eating a little less of the bad side of each move, and across enough trades the little compounds into a real and lopsided gap, in a game whose rules read as perfectly even. the unfairness was not in the rules. it was in the milliseconds, hiding under rules that were fair on paper.
i did not build this because anyone needed it. nobody asked, it is not on a roadmap, it does not ship. i built it because there is a particular satisfaction in turning a vague question into a machine that answers it, where the answer is not a take you can argue with but something that fell out of a world you can hand to someone else and let them run for themselves. that is the most honest form an answer takes. not here is what i think about latency. here is a small universe, here is the one dial, turn it and watch.
i do this more than i probably should. something nags, and instead of looking it up i end up building the thing that settles it. this time it was a stock exchange. next time it will be something else. the impulse never changes and it is not really about the subject. a question you can run is worth more to me than a question you can only discuss, and the surest way to understand something is to build the place where it has nowhere left to hide.