I'd say hardware is, at most, 20% of the speed equation. The rest is technique and literacy. Consider these questions, assuming you're not at risk for RSI or other debilitating problems:
What layout do you use?
How voluminous is your vocabulary?
How quickly and accurately can you read and at what academic levels?
How quickly and accurately can you process the various modes of grammar?
How quickly and accurately can you process words, phrases, and sentences of varying length and complexity?
In addition, add to these all the questions about "What keyboard do you use?" and every sub-question related to that.
There are definitely other questions to ask, more or less specific to each of these. The thing is, there are many ways to improve the brain's pattern recognition of symbols for language learning (our alphabets, words, sentences, punctuation, everything). Practice, practice, practice, obviously, but again, it depends mostly on what kind of foundation for literacy you have. Smarter people aren't simply faster "because they're smart". They're faster because they generally process words better, in more compact, abbreviated, or well-organized forms, for example. More effective pattern recognition can open up the bottleneck to manual dexterity (eye to mind to finger). I mean, you could mash on your keyboard and as long as you weren't typing anything intelligible, you could say that you were typing at 1000 WPM in a not-yet discovered language. Absurd, but my point is, technique is a matter of function. Form follows function. (It's an emergent property of life.) Function is our literacy, form is our physical typing technique.
This leads me to my final question: What are you typing? Have you seen it before or is it completely foreign to you? Colloquial dialogue? College essays? Memoirs? Scientific literature? Speeches? Sermons? Theses?
Starting from the lower end of the spectrum, everyday language is the easiest and fastest to type because they're constantly used and loaded into our short-term memory (some kind of repetition is a recipe for long-term memory). We have many words, phrases, sentences automatically memorized in a sense and we use certain words more than others on an every day basis so typing 100 WPM of ready-to-use words isn't all that impressive if you think about it. If you're talking about transcribing scholarly, graduate-level work in record time, then we get into a strange purist territory. Unless you work for the FBI/CIA in some ultra-elite hacking department, no work is that time-sensitive. In normal life, after 80 WPM, who cares? Anyway, topic-specific language has to be learned, at least for the time that it's relevant. If you don't have this, it doesn't matter how fast you read or how literate you are. It will handicap. I guess my point is that there's still a learning curve dependent upon your intellect and/or linguistic skills. Man, this post is quite ponderous. To tell you the truth, I'm interested keyboards only because I'm fascinated with learning a second language. The PC and its peripheral keyboard has expanded the scope of human literacy. It's amazing what has happened to us since the advent of the printing press almost 1000 years ago. Technology has increased our control not only over our physical lives, but the intellect as well.
My last thought on this is that there's probably some algorithm unique to each person that balances the diminishing returns of literacy, technique, and hardware. Each one has its limit and the weakest, slowest part of the chain will bottleneck the others. I'll accept the possibility that the weakest chain could be technological (who knows, maybe we just haven't found "The Best" layout yet), but apart from mind-reading, there is no faster way to physically process human speech and functional language except through our fingers. Text-to-speech will never match typing because we think much faster than we talk (see the anatomical/linguistic limitations? You could apply this whole analysis to speech itself. The way we breathe, use our mouths, tongues, eyes even; our bodies have their own "machinery" too). Technique will adapt to the technology or vice versa. What's most challenging to improve is literacy. How educated, well-read, well-spoken, or just dedicated to language we are will be the X in this function.
In summary, my conclusion is that typing speed is ultimately the result of combining mind, body, and machine. So simple, but that's what we do.
