Perhaps unreliable is the wrong description. There are significant limitations and these are generally acknowledged. You can do different IQ-Tests and you will

Perhaps unreliable is the wrong description. There are significant limitations and these are generally acknowledged. You can do different IQ-Tests and you will notice a steady increase in your score over time, because you learn about the structure of problems they ask about. Ideally IQ-Tests are systematically different that this would not apply but it isn't possible to create them this way. Your training about similar problems in the past will have significant influence on the score. That does not mean that your intellectual capacity is higher/lower to someone with a different score. Another weakness is the focus on speed or solved problems per time unit. This is also just an estimate and to latest discoveries about intellectual capacity a very rough one. For example there is a corridor of a certain reading speed that correlates with intelligence. But especially at the upper end which also includes more or less average reading speed this correlation vanishes. Same is true with the comparatively easy problems of IQ-Tests. The score might be an estimate, but as it I would call it an unreliable estimate. The limitations of them are intrinsic to the way they work.