Don't Teach Your Enemies To Be Better

From Rest of What I Know
Revision as of 17:31, 4 April 2024 by Roshan (talk | contribs) (Created page with "One of the things people like to do is refine their enemies' tactics. This is usually unintentional, but nonetheless self-defeating. A few examples: * Amazon Fake Reviews: These are usually fantastic since they often point to a low-quality product. "In exchange for an honest and unbiased review, I received this product for free..." was an excellent sign that the product was total garbage. Amazon caught these low-hanging things so now the spam-signals are worse. * Foru...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

One of the things people like to do is refine their enemies' tactics. This is usually unintentional, but nonetheless self-defeating.

A few examples:

  • Amazon Fake Reviews: These are usually fantastic since they often point to a low-quality product. "In exchange for an honest and unbiased review, I received this product for free..." was an excellent sign that the product was total garbage. Amazon caught these low-hanging things so now the spam-signals are worse.
  • Forum Bans: The typical thing is to say "user is banned for talking about A, B, or C" but that just makes it clear where the line is. Modern forum bans are more sophisticated: shadowbans without explanation. This is a good thing. You want your adversaries wasting their resources.
  • Interview cheaters: Pointing out that copying and pasting highlights the text in your shared coding app. "Catching" the cheater is what some people like to do. But doing it well is to not let them know you did it.

In general, when someone gives you strong signal that they're an adversary, you shouldn't discourage that signal because it's a useful detection mechanism for you. You're less likely to turn an adversary into an ally by detecting them and more likely to just make them a more effective adversary, especially if they're a class of adversary that shares information about your detection techniques.

This is obvious to most people where the game is clearly expressed as adversarial, but when it is less so we fail at actions. For instance, we misdiagnose Amazon having lots of fakes as a failure of Amazon. Instead that's the reality of marketplaces, and Amazon is just a mediator. Our adversary is the guy trying to sell us stuff that won't do what we want.