I don't think I'd necessarily call it overused, but it is frequently misunderstood. It's not, as others have said, just an excuse to hand-wave away changes. The technical term for it is "sensitive dependence on initial conditions," and its one of the three properties of chaotic systems. And like just about everything to do with chaos theory, it speaks to probabilities first and foremost.
Which means, in short, you can't just say that because historical figure A was born after a timeline's point of divergence that historical figure A cannot figure into said timeline. On the other hand, what you can do is speak to the probability of historical figure A existing, and the probability that s/he will share certain characteristics with his or her counterpart from OTL. Because of the nature of the phenomenon, it's damnably difficult, if not outright impossible, to be certain of anything. I mean, you can speak in very broad generalizations about the likelihood of major changes emerging, of course: particularly in pre-modern timelines, someone living in western North America is not likely to be as immediately affected by changes in, say, the Roman Republic as someone living in Greece. By the same token, someone born a day after the POD is unlikely to be as strongly affected by someone born fifty years after the POD. That's the basic premise: small changes in initial conditions compound over time in ways that are difficult to predict.
For me, it's an issue of degree. If a timeline incorporates one or two largely unchanged, post-POD elements, I can accept that. If it winds up with more similarities than differences, however, that suggests that the writer hasn't fully considered the implications of the divergence.
Which means, in short, you can't just say that because historical figure A was born after a timeline's point of divergence that historical figure A cannot figure into said timeline. On the other hand, what you can do is speak to the probability of historical figure A existing, and the probability that s/he will share certain characteristics with his or her counterpart from OTL. Because of the nature of the phenomenon, it's damnably difficult, if not outright impossible, to be certain of anything. I mean, you can speak in very broad generalizations about the likelihood of major changes emerging, of course: particularly in pre-modern timelines, someone living in western North America is not likely to be as immediately affected by changes in, say, the Roman Republic as someone living in Greece. By the same token, someone born a day after the POD is unlikely to be as strongly affected by someone born fifty years after the POD. That's the basic premise: small changes in initial conditions compound over time in ways that are difficult to predict.
For me, it's an issue of degree. If a timeline incorporates one or two largely unchanged, post-POD elements, I can accept that. If it winds up with more similarities than differences, however, that suggests that the writer hasn't fully considered the implications of the divergence.