Also all the people saying who keep saying it doesn't make sense to have cheaper, more powerful and smaller engines, because that is exactly what has happened with cars, phones, tv, powerplants, factories, etc. ad nauseam
Technological progression does not eliminate the possibility of design tradeoffs. In many ways, it opens the field for ever larger degrees of specialization. Relative to when the technology was introduced, yes, the average component is probably "smaller, cheaper, faster, more powerful" etc. However, relative to the market, you will generally not see one component which is simultaneously smaller, cheaper, more powerful, and just generally better than everything else on the market. This latter bit is where the specialization technologies fit in (or at least should fit); they're not global improvements from the maturation of the technology, but rather the benefits and penalties you get from the tradeoffs you chose to make.
To provide some concrete examples, let's take computers. There's a single-core 1GHz computer with 512MB RAM and 512MB flash memory (~= hard drive) currently on the market that sells for ~$150. There's an 80MHz single core computer with 512K flash and 128K RAM that sells for ~$100 (no, those numbers are not mistakes; I really do mean 80 megahertz, 512 kilobytes, and 128 kilobytes). Netbooks are available for ~$100+ and have capabilities seemingly superior to either of these. Decent laptops might cost several hundred dollars more. You know what the difference is? The tradeoffs made to get the end product. That 1GHz computer? It's smaller than my thumb, has 802.11G, Bluetooth, a MicroSD card slot, and several physical connections for other components (specifically, a camera and an expansion board), and it's lightweight and low power. That 80MHz computer? It's intended as a teaching and development tool, and has all kinds of peripheral connections you might normally see only one or two of on a normal embedded processor (and for that matter, embedded processors have a tendency to be even more limited in their capabilities, because you're trading off quite a bit of unnecessary capability to make them small, low-power, and cheap).
When talking about specialization, especially in the context of video games where you typically do not get to see everything else that should be going on, it makes plenty of sense for there to be cheap components, powerful components, and efficient components, but not cheap, powerful, and efficient components. It does not matter that cheap components in the year 2225 are smaller, more powerful, and cheaper than those which existed in the year 2200. That aspect of progression is covered by global improvements like Miniaturization technologies and generic industrial output bonus techs. Component specialization is about how you choose to apply your technology - you have a basic part, let's call it A. You can make some tradeoffs to make it smaller (A-I), or cheaper (A-II), or more efficient (A-III). Maybe 20 years from now, the technological descendent of part A (which I will call B ) will be all three of these things by comparison with part A, but you shouldn't be comparing B to A in this manner, what matters is how B-I, B-II, and B-III (the descendents of A-I, A-II, and A-III) compare, and all three of these are still likely to be trading between size, power, efficiency, cost, etc, when compared to their base form B, just like A-I, A-II, and A-III were in comparison to their base form A.