The "it's a proof concept" argument falls down when you realize that automatic contracts would be a serious bug, not a desirable feature, if applied to anything other a wholly arbitrary thingy.
And just as much, the examples are "toys" because crypto-contracts can only work on toys or rather only has it's guarantee properties for things encoded within the blockchain.
The ability to have land change hands automatically with no recourse if said land winds-up in ill-intentioned hands? If you want to trample the rights of tenants, ignore regulations on environmental protection, ignore worries about fraud, assuming all zoning ideas are ridiculous and so-forth, that would be great.
But even if you did want this stuff, it's unlikely crypto could get this because real world things already have their real world keepers because of their usefulness and few are willing to discard the actual protection such keepers have for the mis-features of crypto-contracts.
Edit: Not that I'm a fan of DRM, but it also doesn't seem like blockchains could enforce DRM, which is the primary feature that everyone trying to fence off cyberspace want.
How are baseball cards useful? How about fine art? What about virtual goods in MMORPGs? I guess my point is that enjoyment and fun is a worthy utility in and of itself.
We as engineers or tech oriented are almost too functionally focused and miss the emotional aspects. There is emotional value in all the mentioned goods, mostly in niche groups.
cryptokitties are digital assets that can live on any digital platform like a mobile phone app as an AR filter or inside your favorite video game as a character or a pet for your character. cryptokitties are digital assets that people own. if that isn't important to you, you can't see the future.
For example, they showed that a real-world application instantly cripples the "global distributed decentralized infinitely scalable" blockchain (was it Etherium?).
And it had a fraction of traffic that a real real-world application generates.
Games are useful for ushering new users on to the platform since they rely on novelty instead of network effect. Then when more people are comfortable with using dapps and on the platform, the more serious apps can come about.
All the examples of "toys" he mentions here are also useful. That seems to be the fundamental concept of what grows.
Amazon was useful, you could buy rare books that were previously hard to find.
Ebay was useful, you could buy (and sell) niche collectibles that were previously difficult to trade.
Facebook could show you pictures of classmates and let you communicate with them.
Backrub (ie Google) could help you find information on the early poorly-indexed world wide web.
And so on.
So how are cryptokitties useful?