Security Dilemma
The trap in which one country's defensive precautions look like preparations for attack, so both sides arm and both end up less safe.
In 1898 Germany started building a battle fleet. Admiral Tirpitz wanted prestige and diplomatic leverage and had no plan to invade Britain. But Britain was an island that imported its food, and a German battle fleet in the North Sea could only be read as a threat to its survival. Britain answered with HMS Dreadnought, Germany answered with the Nassau class, and by 1914 both countries had spent fortunes and were more frightened of each other than when they started.
That is the security dilemma: steps a country takes to make itself safer make its neighbors less safe, so they arm too, and everybody ends up worse off. John Herz gave it the name in 1950, and the historian Herbert Butterfield described the same trap at almost the same time.
It exists because there is no authority above states to guarantee anyone's safety, and because intentions are invisible. You can count another country's missiles. You cannot verify what will be in the head of whoever commands them in ten years, so caution says prepare for the worst. Caution on both sides produces a spiral neither side wanted.
What makes it dangerous
Robert Jervis, in a 1978 essay that is still the standard treatment, identified two things that decide how bad the dilemma gets.
The first is whether you can tell offensive and defensive forces apart. Fortifications and coastal guns threaten nobody. Long-range bombers and armor parked near a border threaten everybody, whatever their owner intends. When the two look the same, every defensive measure reads as preparation for attack.
The second is whether attacking or defending has the advantage. When conquest is easy, waiting is dangerous and striking first pays, so crises speed up. That was the situation in 1914, when every general staff believed the side that mobilized first would win. When defense has the advantage, countries can afford to wait and see, and reassurance becomes possible.
Where it shows up now
Missile defense is the dilemma in miniature. A shield is defensive by name and destabilizing in practice, because a country that can absorb a retaliatory strike is a country that might consider striking first. That is why the ABM Treaty limited defenses rather than offenses, and why its collapse mattered.
NATO enlargement, Chinese island bases in the South China Sea and the American response to them, Japanese rearmament, and cyber tools that look exactly like espionage until the moment they go off are all arguments in which each side sincerely describes its own moves as defensive and its rival's as aggressive.
The ways out all involve making intentions visible: inspections and verification, notifying neighbors of exercises, hotlines, and forces deliberately built to look unthreatening, such as missile submarines whose only real use is retaliation after being attacked.
There is a trap in the escape route, though, and Jervis was careful about it. Not every arms buildup is a misunderstanding. Sometimes the other side really does intend to attack, and treating an aggressor as a frightened neighbor is how the 1930s happened. Treating a frightened neighbor as an aggressor is how 1914 happened. From the outside the two situations look nearly identical, which is why deciding which one you are in is the hardest judgment in foreign policy.