The tachyonic antitelephone paradox is, roughly, this:
Alice is in a space station and watches Bob go past at relativistic speed in a spaceship. Some time later she sends him a message at superluminal velocity. Due to time dilation, she expects it to arrive at Bob when Bob has experienced less subjective time than she has when she sent it. Bob then replies - and expects his reply to arrive at Alice when she experienced less subjective time than when he sent it. Since $t_{Asend}>t_{Brecieve}=t_{Bsend}>t_{Arecieve}$ we have Alice receiving Bob's message before she sent hers.
Suppose Charlie is in a space station (at rest relative to Alice) at the point that Bob receives Alice's message. Charlie also receives Alice's message and also sends a response. It should be obvious that Charlie's response arrives at Alice at the same time that Bob's does, but also that Charlie's response arrives after Alice sends her message.
Because the paradox relies on considering Bob in Alice's reference frame and then Alice in Bob's reference frame (but not on the velocity of the message - Charlie and Bob would still get different results if using lightspeed communication) it seems like the core of the issue is changing reference frames.
Now, many people smarter than me have examined this problem and determined that it is a genuine paradox, so what am I missing?