Sunday, March 17, 2013

Two modern banes

A controversy that is boiling with a company of which I have some knowledge (as an outsider) made me think about the modern ways that lead to oops (getting back to the theme of this blog). One can pinpoint exercises related to WWII (or a little later) as the chief starting point for some of these things. There are two in particular with which we can being, namely optimization and lazy evaluation.
  • optimization - Being lazy (next bullet), let me say that it's part of applied mathematics, okay? There is much to discuss here, but let's talk Operations Research, as a field (note 9/18/2013 - incidentally, OR is being used to represent a whole class of disciplines and techniques - must I enumerate the whole mix?). All sorts of system (generic sense, not computational, though there is overlap) problems arose in WWII. Techniques to handle these evolved into the discipline of ORA (an analyst position in the field). What is the bane? As with anything, how well a model (being used lazily to handle a lot, again, okay?) can really fit the world (or predict accurately) depends upon many things. The crux here, though, is that the techniques are taught in school, now. And, that leads to misuse (a major failing about which we ought to care deeply). One really stinking method nowadays is computational modeling overlaid on human activity driving humans as if they were automaton. If you must, recall Steve Jobs' company little thing of suicides by workers who figured out that they had entered hell (albeit, why would a westerner care, it's in the orient?). Also, if you must, let me take you around to show you the same thing in good old America. Methinks (err, knows), some flavor of this reeks from the above-mentioned controversy. You see, many times there was an optimization scheme behind a decision process using incomplete data; the many times can be observed in noting mismatches between what was declared doable and actual accomplishment. It's one thing for you to brag from an armchair (nod to the Madness) about what you can do; it's another to have someone tell you that you ought to be able to do something (without them demonstrating such). "As if by magic" is how some techies have laughed off expectations that were way beyond realizable.  
  • lazy evaluation - Look. Watson (the guy, not the system of late) of IBM, and others, have touted laziness as being the parent of invention, if you would. Figure out an easier way to do something is the usual way to look at it. In computation, AI (as in borgs, et al) made use of this technique. Heck, the web is full of it. Ever notice an iterative push down of data off the web. That is, get back quickly with a little stuff and fill in as things progress. Too, large amounts of data can be partitioned out so as to not be too burdensome. In terms of the above bullet, one can overcome states that are insolvable (computationally beyond any scope that exists now) by using lazy techniques. You, the reader, do it all the time when you rely on other than your own knowledge (look, this applies to the issues of scientism, where everyone has their favorite expert to quote (err, misquote) or makes claims way beyond the bounds of their knowledge). In problem resolution, the lazy approach can help resolve issues, yet it can bring some other issues to the fore. Again, the bane? We have to be lazy (one way to avoid/break analysis paralysis). Yet, we ought to be smart about it. So, let's talk looking forward. It's more rational to talk about re-doing something that has already been done than it is to make prognostications prior to really knowing. Now wait! That applies to data-driven processes, to boot. Why? Extrapolation and false notions of equivalence are a couple of things to discuss here.  
This post is brief, intentionally lazy. But, the theme will continue (actually, we saw it before, quite a bit, as there are many aspects to it, such as out-housing). There are many such banes as these two. Yet, are we teaching the next generation? Oh, like Zuck, and Sandberg, they may not need to know. The problem? As we go along with greater technological prowess, the ties that bind are becoming more problematic.

Remarks:

09/19/2013 -- To some, evidently, grabbing oodles of money, without due consideration of ramifications to others or to the common weal, is the smartest thing; but, we do know that virtue is smart, to boot. Even the secularists are trying hard to show how their worldview can lead to right living (as in, they do not need God to have a conscience). And, what virtue might be prime important to this discussion? Prudence (see Remarks, this day).

09/18/2013 -- About the second bane, engineering cannot cut corners without facing problems. Either there will be recognition of a fault early or there will be a larger failure later. Finance doesn't have this discipline, even in those areas where they claim "engineering" to be a fact for them. See the "misuse" link in the next Remarks which goes back to a Bloomberg Businessweek article about Masters. From the outside, this looks to be classic "laziness" in operation. You see, some types talk as if they work. What they do is work at maximizing their stake while ignoring the issues of Near Zero (I'll admit that I've not described this better - but, intuitively, it's obvious). When finance (and other computation) pays the proper price to do it right (yeah, Zuck, you could learn a thing or two), then we'll have fewer of the stupidity-originated problems.

09/18/2013 --  Pop, fizz, ... Ben had to show largess because of idiots who ran the economy to the ground (rogues all around). Ben is going. What do we have to look forward to? Businessweek has a review issue (of the past five years). Several articles are especially interesting. Too, phrasing shines: spin dross into gold (in relation to mortgage bonds). Perhaps, we'll get back to some of the more pertinent ones, at some point. If we do, it would be to bring forward what has been said here, from the beginning. To wit? Tranche and trash (WSJ has a good take on that). Securitization? This article brings on weeping (one example of the misuse of mathematics and computing that has been harped about).

04/21/2013 -- The view from Saturday.

Modified: 09/19/2013

No comments: