The belief that propagation delay is negligible next to processing time, so distance never has to enter a latency budget. In fibre, light carries about 200 km per millisecond, so a round trip costs roughly one millisecond per hundred kilometres of distance. Any timeout tight enough turns that into a hard geographic boundary: near peers succeed, far peers fail, and the application reports an ordinary failure rather than a map.
Put the round-trip light-speed floor in every latency budget before you tune anything else: about 1 ms per 100 km in fibre. A budget under that floor is not a timeout, it is a distance limit.
Editorial judgements on a 0 to 10 scale, argued in the notes below and nowhere else. They order a list. They are not evidence and they never appear in a claim.
THE TITLE, settled over two passes. It began as a description, delivery does not depend on how far away the recipient is, which is the 500-mile email's observation rearranged. Then Distance is free, which was tighter but read as a claim about money. @moul landed it: on the internet. That is the sentence an engineer would actually say, and the three words that make it unambiguous are the three that name where the belief is held. WHY THIS FRAMING AND NOT THE NARROWER ONE. The first version of this record said that delivery does not depend on distance, which is a restatement of the 500-mile email's observation rather than the belief underneath it. Nobody consciously believes delivery is distance-independent; what engineers actually believe is that physics is never the binding constraint at their scale. Stated that way the rule leaves email entirely: it is the same rule for a cross-region database replica, an RPC deadline, a lock lease, a health check and a game tick. THE NUMBER, which is what makes it a rule. Light does about 300 km per ms in vacuum and about 200 km per ms in fibre, and a request needs a round trip, so the floor is roughly 1 ms per 100 km of separation, before any router, any queue and any work. Paris to New York is about 5800 km, so no amount of tuning puts a round trip under about 58 ms, and real paths are longer than the great circle. WHY IT SURVIVES. The failure never names itself. A timeout under the floor produces an ordinary error, in an ordinary log, at an ordinary rate, and the only clue that geography is involved is that the failures sort by distance, which is not a dimension anyone slices by. SCORES, opinions, argued only here. wtf 8: the behaviour reads as impossible once observed, slightly below the incident itself because the physics is obvious the moment it is said out loud. denial 9: the highest in the catalog, because an engineer will confidently tell you that the speed of light is not their problem, and at a single data centre they are right, which is exactly what makes the belief durable. damage 3: usually a confusing afternoon, occasionally a region quietly excluded from a service for months.