> Like I said, “getting directions” used to be a service. Specifically, back then—and still today, in many places!—you’d be expected to retain the services of a guide when you were in a foreign city/country. Who would often double as your translator, and potentially as your driver as well.

So previously, travel to foreign destinations was something few could afford, since it required sourcing and hiring a local guide. It would seem our current solutions scale much better and at a lower unit cost, opening up the world to more people than could have experienced it before.

Yes, correct. But that’s moving the goal-posts, kind of: “making something available to a wider audience” isn’t the same thing as making something possible. The original claim was that we’re now in a world where these things are possible. But really, we’re “merely” in a world where they’re more widely-available. They’ve been possible for a long time.

What I wanted to highlight, was that technology making something possible, that was previously impossible, is actually quite rare.

There really aren’t all that many innovations that come along and change the world in such a way that a time-traveller from the past, would need to learn an entirely new conceptual framework to understand how we do things now. Almost always, what we do now, maps in an obvious 1:1 way to what we used to do. Hailing a cab? You could hail a cab in Ancient Greece!

In fact, it’s pretty hard to think of genuinely-novel things humans only started doing in the last 100 years, due to some technological enabler. Playing single-player interactive story-games, maybe — even the progenitor of the medium, the Choose-Your-Own-Adventure book, is AFAIK a 20th-century innovation.

The super rich framing isn't useful IMO, much of the advances in technology function to make things possible for people for whom that function was impossible for before.

> You could hail a cab in Ancient Greece!

Depends on who the "you" is. And choosing a super rich person as your point of reference is a pretty arbitrary choice (with underlying assumptions about society that are worth examining, I might add).

The value add is not that the machine guide is better than the human, it’s that the machine guide is available to everyone, at all times.

One might describe technology in general as the way to take the scarce resource of manpower/hired help, and make it plentiful through a narrowing of scope, formalization of the narrowed scope, and then automation of that formalization.

But this misses the finer point of how much is lost in the process of scope-narrowing, formalization, and automation. Or rather, of how losslessly you can truly make such a conversion.

For each well-established technology, it’s interesting to ask this question: given an unlimited budget, is the technology still used? Or is raw manpower used instead? Or some combination of the two?

Truly useful technologies, to me, are the ones that are still used in some capacity even by the ultra-rich, either directly, or because the manpower they hire will themselves use the technology to make their job easier.

A good example of a truly-useful technology is a washing machine. Nobody is hand-washing (cotton) clothes any more, no matter how rich you are. Even if you have a professional laundry service, they’re putting your clothes into a washing machine.

Navigation isn’t quite at that level yet. Your Uber driver uses a GPS auto-nav, but a city guide usually doesn’t, because a city guide is asking a different question — not “how to get there” but rather “which well-known route would the client favour, if I took a few hours to lay out the differences in fine detail.” Which is a question both of subjective inference of the client’s tastes, in a way that would require learned personalization in an automated equivalent; and of a bunch of context factors unique to every city, in a way that makes it hard to reproduce in a narrowed-scope app (rather requiring individual city-by-city coverage, the way only a monopolistic behemoth can achieve.)

So, theoretically possible, but not likely something we’ll see done for a long while, at least until we see some other meta-technological advance (e.g. GPT-4) that makes one or the other part dead simple.