Doing what you know or doing what is right

Often people do what they are used to do, without any reflection. If it has worked once people tend to hope it will work again. Stopping to think and gathering and looking at evidence is often ignored in favor of doing the normal thing. Since conditions will vary, what was done previously may not be the right thing to do.

When you are hired you are asked for experience, measured as number of years working with a given technology, indicating that you are expected to do the same over and over again what you are experienced at, regardless of what the conditions are.

There is also a second kind of experience, where you have seen lots and lots of different things and you know that doing what you are used and expected to is going to go in the wrong direction. There may not always be very much you can do the improve the situation.

A better approach would be to practice an evidence based approach. If there are no evidence, you need to first run an experiment in order to obtain the data needed to make a sound decision.

Efficient architecture

Architecture should not only state how parts relate to each other. Architecture should also describe the forces at work and how the particular tradeoff manifested in the selected architecture has been reached. Further more, information on what options has been discarded and why can be more valuable for understanding the solution than any description of relations.

In my opinnion architecture is about communication of the desired and undesired facets of our IT systems. A obscure arbitrary stack of boxes in a powerpoint picture does not realy provide all that much insight.

Presenting numbers

According to an article in Journal of Extension numbers are understood differently based on how they are presented. Apparently percentages or simple frequencies (1 out of 10) are more intuitively understood than a large absolute frequency (or number like 5 million each year) even though the information is the same. For indicating a small thing an absolute frequency will appear greater in magnitude. Hence an absolute frequency may appear more persuasive.

Personally i prefer using graphs which can convey both kind of information at the same time, as long as the graphs does not get too crowded or too simplistic.

Benefits from ignorance

There seems to be more to the saying "No news is good news" than one would first imagine. In a recent paper titled "Influence through ignorance" two researchers named Isabelle Brocas and Juan D. Carrillo suggest that peoples decisions can be influenced simply by controlling the availability of further information. When no new information becomes available current views will hold, even though absence of recent information should signal caution. The decision maker wishing to influence (or sales man?) can then, when he has the suitable data to verify his view, simply stop investigations generating more data. If generating the data is expensive, complex or in some other way highly regulated, the chances are good that the decision maker/sales man will get away with it.

I guess this can also help explain some reasons why some poorly suitable IT systems are obtained by customers. Since NDA:s (non disclosure agreements) ensure that information is not freely passed on to the public only the message from the vendor is publicly available, and it is most certainly not in the vendors own interest to make negative information available for potential customers. On the buyer side all information can be argued to be valuable, since it gives the rational buyer more sound picture of the product to base the decision on. Some buyers will choose to disregard all information and just go for the decision, like a predator attacking a prey, but nothing can help them. We others can perhaps better try to estimate what the "silence" or "no news" being given us means with this model in mind.

Quantum computing

In the recent issue of Scientific American Scott Aaronson has an article presenting the features of quantum computers. It turns out that although there are certain things that such a computer would be able to do very efficiently it will not be able to solve all NP or NP Complete problems.

Factoring large numbers is however something a quantum computer could do efficiently. The author believes that this will be primarily application of quantum computers will be in physics for making more precise calculations of quantum states. Hope he is right, but i fear that the big monetary rewards for cracking cryptographic secrets used in e-commerce and banking may well be the target in the long run. Once a commercial quantum computer device would reach the market it could render current cryptographically secured data open.

I might be too pessimistic. I remember discussing this with a colleague involved with banking systems at a conference just after reading the paper by Peter Shor around year 2000. My colleague did not think that it posed any real threat although I'm not fully convinced that he ever understood the implications. I would however hope that my e-commerce transactions are kept safe in the future also no matter what devices become available.

Aquiring knowledge outside the box

For many people it seems to be difficult to pay attention to information that is not matching their existing beliefs. Even if you are able to search for anything and find an abundance of correct information on the internet you may simply choose to ignore what is not according to your view.

Researchers from University of New South Wales have found that providing people with the right information on its own does not seem to be enough. As one of the researchers, UNSW Professor Enrico Coiera states: “Even if people read the right material, they are stubborn to changing their views." The study was conducted in the area of health, but i would agree that this is true in most disciplines, especially in IT.

In order to avoid being stubborn we need to constantly question our beliefs, and keep an open mind towards new ideas, but evaluate them in a skeptic way. The real difficult part is however to be able to know what to look for if your beliefs and world view is biased. Only remedy i can think of is to obtain and interact with information from very many disciplines and views, even if the information is of no apparent use, and this way at least become biased in very many ways.

Lerning by imitation

Children seem to learn by imitation according to an article by Derek Lyons et al in PNAS. I would go further and say "So do I", a lot.

When imitating he argues that children cant tell which actions are necessary and which actions are irrelevant to achieving the goal, easily resulting in something he calls overimitation. Seems like there is no skeptic filter screening actions being imitated, although children would otherwise be able to identify irrelevant steps as silly.

I guess this occur a lot in IT (among other disciplines) as well. Part of what is done has no real relevance to the problem at hand. The procedures just happen that way as a result of accidents being replicated and accidentally frozen behavior, just because someone happened to originally do it that way. Somebody may finally realize that some parts of procedures are not bringing any value and improve, but that seems to take ages.