The 'Just Evil Enough' Ethics Test
An Occam's-razor line between subversive-and-clever and actually evil.
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 90%
Because subversive tactics border on manipulation, Croll devotes a chapter ('Don't actually be evil') to a bright-line filter. Before running any exploit, check it against five hard 'don'ts'; if it violates one, it's too evil regardless of how clever it is. The recurring self-test is simply 'was it evil? no. was it clever? yes.'
Origin
Formalized by Alistair Croll and Emily Ross in 'Just Evil Enough' as an explicit ethical guardrail chapter, drawing a deliberate Occam's-razor line.
Core principles
- 01Subversive means using a system unexpectedly, which is clever, not evil
- 02The test to pass repeatedly: was it evil (no) and was it clever (yes)
- 03Being 'just evil enough' is a sliver of unexpected behavior, never harm
How to run it
- 1
Never assume consent or act without permission
Don't opt people in, extort, or take an action on someone's behalf they didn't approve. Get Satisfaction crawling complaints to extort brands, and Afterlife crawling obituaries (sending funeral homes flowers years later), crossed this line and drew criminal charges.
Watch out Assuming consent is the most common way a clever aggregation or bait-and-switch tips into evil.
- 2
Don't lie or use dark patterns
Deception and manipulative UI (like exit-intent tricks that harm the user) are out, even when they'd work.
- 3
Don't break the law or its intent
Includes the spirit of the law, not just the letter; a legal-but-clearly-harmful move still fails.
Pro tip Loopholes (parklets, the PT Cruiser counting as a truck for fuel standards) are fair game precisely because they're legal and harm no one.
- 4
Don't punch down
Direct subversive energy at those in power and the status quo, never at the vulnerable.
- 5
Protect your reputation
Ruining your own reputation is usually a bad outcome; weigh whether the tactic survives daylight.
Pro tip Validate the offer instead: for bait-and-switch, ensure people are genuinely excited about what you switched them to so almost none complain.
In the wild
Localmind found 95% of strangers would answer geofenced questions, but few users asked questions. Croll suggested seeding the platform by posting questions themselves so users could see what answering felt like. Rachitsky worried this was evil; Croll's line was 'I'm not saying be evil, I'm saying be just evil enough.'
→ Seeding modeled the desired behavior (like Scott Belsky borrowing designers' templates for Behance) and helped solve the cold-start problem without deceiving anyone.
Common mistakes
Believing a just cause speaks for itself
People on the 'right side of history' often let their cause speak for itself instead of fighting for it, and lose to those willing to bend the rules.
Being too much of a Boy Scout
Excess conscientiousness stops founders from making legitimate, non-harmful subversive moves that would prove their idea is better.
Is it for you?
Best for
Founders about to run a subversive or growth tactic who need a fast go/no-go check on whether it's clever or genuinely harmful.
Not ideal for
Teams looking for permission to manipulate users; the test exists to forbid exactly that.
From the transcript
“don't ever assume consent don't act that permission don't punch down um don't break the actual law don't use dark patterns”
“there are certain things like abuse like assuming consent and acting without approval uh lying or using dark patterns breaking the law or even the…”
“I'm not saying be evil I'm saying be just evil enough”
“was it evil no was it clever yeah”
From the episode
Just evil enough: Subversive marketing strategies for startups
Alistair Croll (author, advisor, entrepreneur)