It's absolutely not "bullshit" to develop games that way.
Both ways have advantages and disadvantages; the "horizontal" method lends itself well to small incremental improvements and utilising as much of the team at a time as possible, whereas the "vertical" method allows experimentation of different things to help build a picture of what the full game should be like - for example, if there are fundamental problems with the integration of gameplay and art style changes can be made on the small portion of the game with minimal pain.
The former approach is what you do if you're absolutely sure of what you're doing in advance. It's ploughing your whole field at once, sowing all the seeds at once, watering all the crops at once, to get a full harvest at once. You can adjust the seeding as a result of the ploughing, or the watering as a result of the seeding, but generally speaking you'll have a plan and stick to it.
The latter approach is what you do if you're trying things out. It's a new crop, so you're not sure how deep to plough, how much to seed, or how much to water, so you take a small portion of your field and do everything at once to that small portion and see the result (the analogy breaks down because plants take time to grow, but just assume they don't). If the ploughing wasn't adequate, you haven't wasted time ploughing the whole field - you've only done a small portion.
The other thing is that pitch demos are important to most games. The Shenmue Kickstarter sold on name alone - they could've had absolutely nothing to show for the game and people still would've thrown money at it. This isn't the case for most games (crowdfunded or publisher-funded alike) - people want to see what they're backing, they want something representative of the finished project, because otherwise you're selling them an idea and not a product.
