In RISC the basic idea is a lot of little CPU instructions that can run through the processor at a faster and more predictable rate. With CISC the idea is many different forms of complex instructions can be used to do exactly what's needed in a single execution. Now, I'm not here to debate what's better, RISC or CISC, they each have their own merits. The analogy between instruction sets and RDF centers on the idea of being 'bit stingy'.

When RISC computing first emerged (and again, I'm not quibbling over timelines) there were great limitations in disk space and network throughput. RISC programs, by their nature, tend to be a good bit larger than those compiled for a CISC chip. Again, there are ways around this but the basic premise holds true. As a result many people bemoaned using RISC because it would waste their precious disk space. We're talking average hard drives being 40 megabytes, so this wasn't an unreasonable complaint. Likewise, ethernet wasn't the pervasive thing we have today and modem links between sites were far more prevalent than actual high-speed lines. So for something to make use of RISC it would incur resource costs. The people siding on the CISC camp vigorously assailed this as being a flaw of RISC. Of course, it wasn't, but given the conditons at the time it was easy to take potshots at it.

Fast forward to today and the 120 gigabyte drive on my local desktop, the terabyte I've got on the server side and a high-speed DSL line suck the air right out of those arguments.

So what's this got to do with RDF? In RISC it's possible to run things a lot faster but it requires using simple steps. The value is in being able to run those simple steps a lot faster than is often seen on CISC. When marking up data in XML it's possible to use really specilized markup very narrowly suited to the task. This is like CISC. Conversely it's possible to use very simple markup like RDF. But like RISC, more markup is needed to make the point.

Where things start to make a difference is in making use of a lot of that data. When it's marked up in RDF there's a better chance of using much more RISC-like processing to search and retrieve the data. Using complex markup means much more complex search and retrieval is going to be needed.

So what's better here? Simple markup that can be a bit verbose but is much more versatile or compact but complex markup that's very difficult to process later on? Smart folks will recognize it's the former. Yes, stuff marked up with RDF/XML style, following the RDF model can be more verbose. But by following the model the later processing stands the opportunity to be significanty easier to process in bulk.

There's more to this idea but I just wanted to get the analogy out there. Is it worth being bit stingy in the use of complex XML when it means much greater hassles in the long run?