Memel

2. Unoptimal algorithms (...continued)
KISS me...

In order to understand this my proposition, lets determine what is it, in my means, - the infinite (or unlimited) power of computer system. Yes, sure, - it is a computer system of unlimited resources and speed. Yes, I know that it cannot exist at any times, never ever. So, I'm forced now slightly to deterpolate this definition towards more possible and normal situation. How much power do you need today? How much do you have now? Add 200%. Not enough? Add 1000%. Quite enough? This means, that the today's "infinite" power is, in fact, reachable. And even more - it is not even strictly necessary to reach that power in order to start using the new programming and system design concepts that I'm going to talk about.

Thus, herewith we defined the terminology and the concept of "infinite power".

Now we are in "infinite power" world, and we can start wondering how we can live inside it. In other words - how we can take advantage over this situation?

Lets take a simple common situation. Databases and indexes. Since we are "unlimited" (in resources and speed), lets think - whether we must continue using binary and sometimes even trinary trees for indexing and searching for information? Or can we just pass through linear text file to search the pattern? If this sounds too terrible for you, - Ok, lets take a step back and create indexes. But should we use the binary structure for them anymore? In the world, where computing power gets close to unlimited, there is no need of this excessive technology anymore. Any other "unoptimized" algorithm provides the same result in speed and functionality. More than that - it is much simpler and more understandable, thus - manageable and supportable. So, in this world, old indexing practices becomes kinda forgettible technology (technology to forget).

We have a plenty of such "forgettible" technologies nowadays. Just remember the nightmares of VMM (virtual memory management), memory locking, task switching, multithreaded filesystem access (together with swapping/paging) and so on, and so on...

And here I'm coming to another part of my proposal concerning the culture of writing mordern and future systems! The KISS principle is not enough! In fact, this principle is more about writing the most possible simple code in our present/past situations, but we need to transform it for writing future, "unlimited" applications/systems. It took about 30 years in computing history to invent rather logic - WYSIWYG interface paradigm. It was nearly impossible to invent it in starting years (40-60ies). Think why? Correct. The power of computing system had to become adequate for this to happen. And now - another 30 years passed since that invention. And now I'm dare to introduce another shift in computing psychology. This principle is, shall I call it:


3. What You Write is What You (really) Think
I don't expect you'll understand me...
(c) Linus Torvalds

Did you ever have a brief look at the Linux kernel code? I cannot ask you to look inside the MS Windows code, unfortunately. But I do expect the similar situation. Guess, what percent of comments on these code-chunks is really about the normal flow of things? In other words - about the normal flow of normal people's mind when they are in situation of examining problems? I'd count only for 10-15%. The rest and the biggest part of comments are about "enormal" thinking. And this was quite normal when we spoke about programming all the past days. Every programmer suffer from it, and that's why normal programmers doesn't like to write comments. Ones he had "invented" pervasive way for solving the problem, he wouldn't like ones and again explain it in details! For example, when he takes sorting as a problem, he doesn't dare to solve it in normal flow of his mind! Because the result will be slow. Thus, he forced to undertake his mind and to think, how to do it in more sophisticated way, only to make it work fast. And only few programmers, sometimes, gets down to the Earth, and lets into describing this "undertaken" mind and write details of this thinking as an algorithm.

D.Knuth is the most notable example of this - in "par excellense". Besides his undisputable talent, many, and in near future, - most of his hard work will nicely fit to our "forgettible things" category. I want to excuse for my such categoric position. Of course, the humanitarian values of his works are and always will be significant. I speak here only about the ultimate thinking princilple of the whole past epoch of programming.

Surprisingly - some things, which were "Wrong" in the past, suddenly becomes "Right" in the future! I'd like to relax you - that was happening always. Back to my days when I worked for one bank, we had a very nice "joke algorithm". We called it "Stepas's algorithm". Who of us would dare that time even to think, that it can become quite good in some future! The algorithm itself was nothing special, except that it made several passes through the database table in order to print totals for some selected accounts. When launched it was impression, that the whole computer system had been hung, except that sometimes it printed a line of a total for one of the accounts. Yes, we asked Stepas to rewrite this program to make one and only one pass through the account table. And, since then, it was a continuous enjoy about this algorithm perspectives and further abilities.

But now, when I think about Stepas's algorithm, I become more serious. Why on the Earth, have we to exploit our mind in such erodic way, only in favour of underpowered computer, which will never be such underpowered anymore! Why not to think in a straight (let me say - naive) way, for it to make things simpler, more understandable, and more manageable?
Why go opposite direction - away from our [Nn]ature?
I'm not to tell, that from now we all should write all our programs in Stepas way. I'm only to tell, that some "incorrect" algorithms suddenly may become correct in some future, especially when we speak in the terms of the task goal, not of the performance.

to be continued...

Memel is the old name of Klaipėda city. It resides on the east cost of Baltic Sea, and now belongs to Lithuania. Here is Wikipedia entry for Memel (Klaipėda).
I also chose Memel name for my Infinite Power Computing Theory Project, the basics of which I'd like to present here.