"It's all contextual and relative, which means the same model that didn't work last year may work this year because of how the situation has changed. But nobody knows that until it actually happens."
"In fact your rhetorical questions I quoted above--which you probably meant cynically--are very interesting questions and I encourage that you seriously think them again. You seem to think they're idiotic but I think they are brilliant questions."
I agree. The point I'm making is you can sometimes know these things because they were obvious with 20/20 hindsight & market feedback. The questions weren't cynical so much as best examples I could squeeze out of my brain in a hurry that I could justify with examples. Glad you liked them. :) Let's look at them.
First, Gopherspace is gone in about every way you can think of. I predicted it could return on feature phones with a new implementation. There's a system in developing world that's essentially that & text-based. Yet, it's an inferior experience to both Web and native due to lack of rich content and network effects. Almost all growth happens due to network effects. So, deny the network effects, then you're almost guaranteeing no growth. An imitation of or improvement in Gopher model will fail unless it addresses this. Every company that I ever heard of that tried failed. Of course, modifying Gopher to have network effects basically turns it into Web 1.0. ;)
Second example is paying a lot for what they can get free. The exception where this works is luxury or coolness markets. I'm ignoring that here, which I should've mentioned, because it's the exception. You release an email client for personal, not business, users that cost $50 a month... you will fail. Even high-assurance security, my subfield, had a really hard time pushing messaging and email systems that baked in strong, endpoint and protocol security due to premiums charged for high cost. We saw same for usability, platform integration, etc. It had to have certain features and not be too much money. Even Apple, an outlier, went through hell trying to fight this rule until they acquired Next Computers and their I.P. that was worth extra money by being substantially different and cool. Even iPhone follows my rule because they had to basically put a Mac in a phone & hit coolness crowd to charge what they did. Try this stuff with a basic differentiator and too large a value? It will fail. Companies keep trying and failing on that.
Third example: "Do a mobile solution that requires a specific device rather than smartphone?"
Pagers. Mobile communicators. Briefcases for secure, satellite phones. Ultra-thin, netbook-like devices. GPS systems. PDA's. Tomagatchis. All sorts of things came before cellphones got good enough to do the same stuff with more convenience, low weight, etc. People kept trying to market the same kinds of stuff from there with 100(?) failure rate. Question mark as something successful might have slipped past me that wasn't an outlier. Browsers, productivity apps, note taking, telling time, games... all of that goes on mobile outside extremely nich markets (eg Garmin's). Smartwatch's might be an outlier that cleverly exploits that a watch is only other thing that's acceptable and semi-mandatory to carry. It doesn't break my rule: it replaces an existing, marketable category of product with added value. Otherwise, virtually everything like before is failing if it's not on a phone. Next one will too without compelling reason.
So, it's not about experts or limited views. Even an average person will tell you they don't want to carry a 2nd device for your note-taking app. They want a pretty Web interface or good native one over manually navigating Gopher folders. That the app isn't worth a bunch of money if something else does almost same thing for free (also see "Worse is Better" for 90% rule). In all of these cases, someone should've stopped to interject these issues to steer the startup toward an offering that lacks these risks and has components of success. After all, the ones that succeeded mostly had those components and we're play the odds where we can right?
I'm for experimentation and risk-taking where context changes. It's just that many attributes keep showing up in both successful and failed products. So, making one or avoiding the other could benefit from those most of the time. That's all I'm saying. Past that, a proven thing that applied in the past might suddenly apply again in new circumstances seen only by a visionary. Go for it. Try it. After all, I am "that guy" (or one of them) on Hacker News that keeps reminding IT people they're always re-inventing stuff (good & bad) from the past without learning lessons from those that did it right. ;)