Because the stakes are so low

Many disputes over academic credit involve exchange of niceties, scholars “thanking” each other, claiming this isn’t really important, etc.

But in fact it appears that it is important.  Why?  Perhaps because you spent much of your life on research, for a relatively meager salary and at a substantial personal cost; your main reward has been your ideas, which you have nurtured for years, decades like babies, and now you can’t stand that they are mistreated. I am inclined to think that some of these issues are especially tricky in theoretical computer science, since the field abounds with interrelated concepts and objects (as opposed to raw true-false claims).

In my experience with a little effort and communication one can come up with wording that makes all parties happy enough.  I list some practices that I think are good next.  But because of the lure of negative thinking, I list them as malpractices.  The reader will spare me the irony of the mathematician who after exerting tremendous mental effort to concoct the subtlest argument is compelled to spell out the contrapositive of their theorems.

1. Cut the communication.  You spent years writing the paper, and now out of the blue of your mind this other work shows up to ruin the party?  You are receiving emails from a stranger?  Sorry, enough is enough.
2. Delay updates as much as possible, if forced to make them.  Especially miss the camera-ready deadline, blame the tight time interval, traveling, newly discovered apple allergy, anything.  That camera-ready version will be available forever, tough luck!  Delay uploading revisions to the online repositories as much as possible, to allow your version to seep through the internet engines, creating a sediment that will be nearly impossible to remove.  Or delay indefinitely, promising updates you’ll never do.
3. Don’t show your revision to the other party before submitting it.  Keep them in the dark, keep them clicking.
4. Make no mention of the issue in the abstract.  This already gets you 90% of the readers on your side, maybe 95%.  Discuss the issue as late as possible in the paper, the effect of this practice is magnified if your paper is already very long.  I’d say each page gets 90% of what’s left of the audience on your side.  Pretend this doesn’t matter because people don’t read papers sequentially anyway.
5. Don’t state facts, instead use careful words or expressions that can steer the reader’s opinion your way (like implicit, implies, for example, e.g.).
6. Be overly fussy about planting flags: If there isn’t a boldface claim with parameters exactly matching yours, it isn’t there.
7. If forced to cite X, find a more interesting paper Y to cite before or in conjunction with X.