Tarpit Ideas
There's a class of ideas that founders tend to gravitate towards. These ideas pass the basic criteria for idea generation – they are usually interesting (and seemingly doable) to a wide variety of potential founders, users claim that they want this feature or app (via user interviews or other feedback), and there's something emotional about it.
Dalton Caldwell and Michael Seibel at Y Combinator call these ideas "Tarpit Ideas" (a much better term, much better than my "Flying-Car Syndrome"). There's probably no other institution that has seen more nascent startup pitches than Y Combinator, so they are in a unique position to think about this. Some notes on some of the ideas they touch upon in their conversation, "Avoid These Tempting Startup Ideas."
Consumer ideas – attractive because they usually have much bigger outcomes and are more visible (we all use Google, Instagram, Apple, etc.). Very reliant on timing (point in the S-curve for broadband, mobile phones, etc.). The bar is also very high: the best products need to be immensely stuck and obvious without any marketing or branding.
"App to discover new things" – Discovery problems are usually not real for a variety of reasons: the underlying supply doesn't actually exist (people have trouble picking a restaurant to order food from, but that's because the options are quite finite), or the demand doesn't actually exist (i.e., most people like popular music or popular restaurants).
"Ways to bet on things"
"Rebuilding everything in Web3"
Those were the ideas that they listed in the conversation, but I'm sure there are many more. Some of my own ideas:
"Craigslist killer"
"Deploy your software with one button"
"Write any app without code"
"A new Email protocol"
Virtual assistants / universal chatbot (2017)
"Connect any service to any other service"
Super app (US)
It's not that these ideas are inherently wrong or bad, but they require an extra layer of thinking for the questions: why now? why this team?