One of those vids mentioned equipping weapons onto vehicles too, suggesting SilkWorm and similar are doable. The nature of the beast is actually very different than it first appeared, and I wonder if this is the nature of the future? That is, the game was unveiled as what see as a platformer with customization. We all wondered what extras would appear like 'baddies'. The devs have added all this gubbins, and spoken of aiming not for a 'platform game' but a 'game creation engine'. I think their intention from the very beginning was to open up game development to everyone (I think there's an interview out there saying just this). The stuff we're seeing, the features at least beyond the clearly typified levels that form a cohesive feel, are offering quite an array of possibilities.
How does this affect development and point to the future? 20 years ago, to write a simple game of Manic Minor required complex understanding of computers and coding. 10 years ago, tools like AMOS, Blitz etc. provided high-level access, so those same games could be created with a fraction of the understanding and effort (and actually look better!), but of course still needed a specialised understanding. LBP offers the same opportunity to create those same game ideas, but without the needing for programming know-how. It's not just a game, but a development platform. Creating games has progressed from punching in hexadecimal numbers to writing assembler to writing C to using higher level languages to construction engines like SEUCK to LBP, the highest level of game engine!
20 years on from now, will technology be sufficiently advanced that an engine can exist where the user can throw in objects, associations, animations from a big library, and recreate the games we have now without having to worry about how it works? Will the future of game development tend towards complex engines offering a simple WYSIWYG (or should that be WYSIWYP?) interfaces? If UE3 can handle an assortment of games, how much work is needed to create an interactive engine interface?
The way LBP is structured seems to suggest this is indeed possible. Standard mechanics of rendering, physics, materials etc. are generic across titles, based on the real world, so sharing those features within an engine across all titles makes sense. That would leave the developer to worry themselves solely with content. It wouldn't take much adaptation to create a vertical/top-down variation on LBP, nor to add player functions in the form of equpiable tools activate by a keypress. Then Bionic Commando, Commando, 1942, and everything else is doable in the one engine. The engine is limited by technology, but as technology advances those limits can be shifted. If we had enough power to simulate the real world with exactness, and just deal in materials and structures, leaving the engine to determine clouds of dust from wheels, damage from bullets and rays, ambient lighting etc., we could just create content.
So I wonder where it'll go and how empowered Joe Public will eventually become.