A runway has to meet two basic criteria, bearing capacity and sufficient length. Where the first one rarely is a problem, the second one is. Several times each month an airliner finds itself in desperate need for a few hundred extra feet of paved surface. The lack of that precious margin sometimes cost passenger lives. But not that often. The aircraft involved are on the other hand, more often than not, a total write-off, sometimes by post-crash fire and unfriendly terrain.
Take-off runways and landing runways are obviously the same thing, depending on usage. When the length is not sufficient for the full structural weight of the aircraft, the weight has to be reduced. Take-off weight- and landing weight calculations take care of that. If there was a guarantee that all systems functioned and all engines were always running, much higher weights would be possible. That can of course not be taken for granted, and since the industry has margins for almost everything, take-off and landings are no exception.
During take-off the basic idea is that one engine shall be allowed to fail at the most critical moment without causing a catastrophe. Based on the actual weight of the aircraft, quite often not the full weight, but the max calculated weight based on criteria such as temperature, wind, barometric pressure, braking action (=how slippery is the runway) ets., but disturbingly often also by the lack of sufficient runway length, a critical speed called V1 was introduced many years ago. If an engine fails at that speed, the idea is that you shall be able to either stop with maximum braking at the runway end (now there is suddenly no margin), or continue the take-off and reach a hight of a few feet (normally 35 ft, not much margin there either) over the runway end. Should the failure occur before or after that V1 speed, you obviously have less problem – and more margin.
Since engines almost never fail, and even more rarely at the exact moment the speed has reached V1, nobody worries very much. Those who have tried to stop at V1 has found out that the calculation was more wishful thinking (not taking into account unused runway behind the aircraft when lining up, worn brakes, reaction time etc.) than based on actual ability to stop. Thus the general recommendation to prefer a continued take-off rather than an abort.
During landing the means to stop the aircraft are wheel-braking and engine reverse. Engine reverse capability gives extra margin in the landing weight calculation. Buckets are deployed behind the engines exhaust, directing the power forward and more important killing the normal thrust. (The efficiency of reverse thrust is not that high, but actually high enough to enable smaller aircraft to back up, should they need to.) The maximum landing weight is then calculated like for take-off, depending on the same criteria, where braking action and – again – runway length are crucial. Then there is a margin regarding percent of runway remaining once the aircraft has come to a complete stop. That margin has a tendency to be reduced at places where no one is interested to pay for a runway extension. Sometimes terrain around the airport prohibit longer runways. Thus the rather cute comment on a pilot landing chart for an airport in Sweden: ”If over-running the runway, turn right to avoid the downslope towards the river”
The question remains why airports have not been extended when higher aircraft weight warrant it. The is normally room for it. Often there is a lack of know-how among those who are in charge of airport developments. Know-how as to things like what a B747 at full landing weight need if the braking action is poor (or nil). The cost is not enormous. Not compared to the cost of lives and totally wrecked aircraft. And taken into account the gigantic loss of revenue around the world where airlines cannot use their big and expensive aircraft to their max capacity, there is an answer right there. The industry has moved forward from the propeller era. Many airports have not.