Nov 10, 2006

Bongi

This week my brother started his own blog about being a surgeon in the notorious Mpumalanga province in South Africa. Three rich posts in only the first week make my own blog-rate of about 3 a year look rather dismal. Also the content is very interesting, dealing with the hard issues of facing death in an environment that seriously increases the chances of death by violence (read the blog for details!).

And his text is very well written, to the extent that I'm almost able to ignore the strict lack of capitalization :-)

Oct 6, 2006

Google GapMinder - Wow!

I recently read a blog posting by Simon Chapman called simply Google Gapminder Beta 2. The google application it references, Gapminder, is absolutely fascinating. It is an AJAX style web animation of global human population statistics that are most incredible to observe.

The original Gapminder data with visualization tools and presentations is available at http://www.gapminder.org, but I spent more time running the google animations, and here are a few points I found interesting:

  • phone versus internetAnimating phone use versus internet use shows a slow increase of phone use over the last 30 years, with high use countries remaining high, and not much shuffling, and then over the last 15 years internet use races to catch up with phone use, mostly reaching it, with individual countries increasing internet use in growth spurts. Now, in most cases, internet use and phone use are highly correlated, or even matching. In the late 90's, not surprisingly the Scandinavian countries, Sweden, Norway and Finland lead the pack, with the US close on their heels (and the rest of Europe and richer SE Asia a little further back), but over the last 10 years, possibly drive by internet use, the phone use seems to have sped up, with the global distribution narrowing, and all of the wealthier countries crowding up in the maximum penetration area.
  • life expectancy versus incomeAnimating income versus life expectancy is even more amazing. For the first 15 years there is reasonable global correlation, with low income countries having low life expectancy, but in the last 15 years most of Africa has veered severely off that correlation line. Consider Botswana in 1990, with the highest income of continental Africa, sitting right up there with the better off third world countries in Asia and South America, on both income and longevity. Then over the last 15 years, while income continues to increase, the life expectancy drops from 65 to 35 years. This was the most dramatic impact of AIDS I had ever seen. We do not feel this in Europe, but certainly it is horrific in Africa. Both my brothers are doctors in South Africa, where life expectancy dropped from 63 to 45 since 1992. They have often described the severe impact of AIDS on their work, and like most doctors in South Africa are frustrated by the authorities apparent lack of interest in helping the situation.
  • On the same life-expectancy animation above, you can see Rwanda hanging around the 45 year mark in the early and mid-80's, but in the late 80's it dramatically drops, to an incredible minimum of 24 years in 1992 due to the civil war (read 'genocide'). This atrocity was so horrific it makes Bosnia look mild, but being in deepest, darkest Africa did not generate the same media coverage as the European conflict did. Since that date, the life expectancy has come back up to 44, but income has slightly reduced (with a big drop in the mid-90's from which it has partly recovered). Africa is surely full of the most amazing, but unfortunately horrific patterns on this gapminder plot.
  • contraception versus childrenPlotting female contraceptive use versus children per woman is also interesting. Not surprisingly there is a strong correlation, with high contraceptive use linked to fewer children. China sits in a strong position, with a very high level of contraceptive use, peaking at 90% in the mid-90's (and slightly less than 2 children), and most of Africa sits at the other end of the scale, generally below 20% and around 6 children (South Africa is unusual at 56% and 3 children, but still on the correlation curve). One fascinating anomaly is Russia, remaining consistently below the curve, with surprisingly few children considering the low use of contraception. One wonders if there is a cultural aversion to admitting contraceptive use, or perhaps just a low incidence of sex?

Wow! I could spend hours poring over this incredible mine of information. But I better save the rest for another day :-)

Mar 29, 2006

The perception of control

Over the last few years I've begun using the phrase "the perception of control" to describe a phenomenon in company decision making I've seen unpleasantly often. I view it somewhat as a successor to my previous pet phrase "un/informed decision making" and coupled to the phrase "the illusion of efficiency" which I’ve also enjoyed using.

Un/informed decision making

I'm pleased to say I've spent most of my career working for small startup companies, where necessity requires that decision making is done by the same people that actually do the work. This usually leads to relatively well informed decision making. However, at least once I have worked for a large company that was structured with vertical silos such that decisions were made by managers that usually knew very little about the subject on which they needed to take the decision. Generally they also didn't have the time or inclination to educate themselves, or at least talk to the 'people on the ground' who actually knew something. Take a look at a recent posting on this subject by an IT professional forced to follow IT decisions taken by his IT-illiterate boss: When servers crash and burn

This vertical silo structure has a secondary effect when large companies try to increase efficiency by creating narrow, specialized roles, which leads me to my second phrase:

The illusion of efficiency

Back in the same large company, I had the rather illuminating experience of being part of a 5 man international team that took 6 months to install a printer. Yes, you heard me right, six months and 4 teleconferences simply to install a printer. This amazing level of extreme inefficiency was the direct result of IT projects designed to increase company operational efficiency through specialization. The point was that if each individual only did one specific task, they would do it faster, perhaps as much as 20% or even 30% faster than someone switching context between several tasks. The end result: IT support was located in Asia, core networking in Germany, Windows domain management in Norway, hardware requisition in Belgium and Project management in Scotland. With business management in California, things could have been even more complicated, but luckily they were only required in one of the meetings. Had we had an onsite IT individual handling all the various IT-related tasks, the project would have taken a couple of hours at most.

More recently I've been in discussions with my current management on the subject of software development efficiency. They can see that 20% to 30% efficiency gains might be possible through the reuse of software code components across the company product lines. I have argued that any fractional operational cost gains will be offset by dramatically increased time-to-market losses. We're not talking 6 months for a 2 hour project, like my previous example, but we're certainly talking about taking 6 month software projects and scaling them up to years. This problem is nothing new. Software companies have battled with these issues for decades, and many books have been written on the subject. One of the earlier ones is ‘The mythical man-month’ covering related issues, but more recently the flood of Agile development books, like ‘Lean software development’, ‘Extreme programming’ and of course the entire ‘pragmatic programmer’ series.

Drawing from experience I can say that for years I too believed in trying to increase software development costs through increased code re-use, or the sibling concept:- coupling existing software projects together to prevent re-inventing the wheel. However, retrospective analyses of such projects lead me to a few ‘startling’ conclusions:

  • Projects easily became much more complex than initially expected if any one of a few conditions existed:
    • more than one pre-existing code-base or product
    • more than one team of software developers
    • specification and core development are geographically separated
  • Projects intended as ‘prototypes’ or ‘research only’, and as a result were run with small teams with close access to a ‘trial customer’ magically hit much, much tighter deadlines and often with a feature set and level of quality akin to what would have been planned with a ‘real product’ development, but minus much of the heavy project planning.

These observations have made me a natural believer in much of the new Agile development methodology that is such a hot topic these days. Clearly there is a growing body of software developers that no longer believe in ‘silo’d’ projects with ‘uninformed decision making’ and ‘the illusion of efficiency’ driving project decisions. I’m certainly one of them, but unfortunately many managers are not in agreement. This leads me to the main point of this article:

The perception of control

In a small startup company, where there are few people interacting, people generally have a high level of ‘control’ over their environment, colleagues or projects. This is a natural consequence of the fact that any decision maker needs to talk to very few people, if any, to take an “informed decision”. Quite often the decision maker is actually the person that knows the most, and the person that needs to follow through with actions. However, as companies grow, the number of employees increases and jobs become more specialized, decision makers start ending up in the position where they no longer know enough to make an informed decision. Things can go in two ways:

  • Delegate. The decision makers entrust aspects of the decision to others who are closer to the data, the customers, or the people on the ground. This requires the ability to trust others, to stick with decisions and to not change the rules of interaction or balance of authority depending on the current political climate. It is a challenging task surprisingly often excelled at by people who are not entrepreneurs, and need to delegate and trust by necessity. Coupled to this is the ability to take blame when necessary, and not look for scapegoats. See this very interesting article on the subject: My Bad!
  • Dictate. The decision maker takes the decisions themselves, and maintains total control. This is where the ‘perception of control’ comes into play. No matter how brilliant any leader is, they can only scale so far. Sooner or later a growing company exceeds their ability to maintain genuine control over everything that happens. This is a very common situation for entrepreneurs to end up in. The same strong leadership characteristics that helped that person excel at starting a company and build a new business, often lead to this situation where they cannot successfully delegate. All decisions end up being made based on the dictators ‘perception of control’. They can never have real control because the company has grown beyond that, however their need to feel in control means that they will take decisions based on that feeling, not on real knowledge of the situation.

Paradoxically enough, those that are most dictatorial are exposed soonest and often thrown out by the first controlling VC around the corner, while those that try hardest to do it right and delegate are those that end up maintaining the dictatorial approach for longest. But it cannot last. Dictators are always overthrown.

So, how do you tell the difference between true delegation and dictators operating with the ‘perception of control’? Well it’s not as hard as you might expect. There are a few rather simple things to look for:

  • Middle managers admit that success within the company requires having the ear of the ‘decision makers’ (ie. decisions are made through ‘lobbying’, not informed decision making). I observed this situation once when an external management consultant asked key questions as part of management training, and was somewhat distressed by the answers, even though several of the managers involved did not see the problem (since they did indeed have the ‘ear of the decision maker’).
  • Strategy decisions are taken in closed meetings, or without the considered input of a reasonable representation of management. This one can be harder to judge, but can show itself through a simply polling of people in the company with the question: do you feel that your opinions are considered in company decision making? Another side to this situation is that the people placed in specific decision making positions end up not having real decision making power, and become mere puppets, which is usually a very unsatisfying role, leading to lack of motivation and often early departure from the company.
  • The use of FUD.
  • The decision makers are never willing to admit they are wrong: See My Bad!
  • Individuals that express contrary opinions to the decision makers are removed from positions of influence. When such people are few, this might not be a real sign of a problem, but when this involves a large number of people, you know immediately that it is a major effort to maintain ‘the perception of control’.
This last point reminds me of a situation I experienced a few years ago, where the large majority of the people involved in a key project were removed from the project and from positions of influence, in order to kill the project. This drastic level of ‘surgery’ was required because the particular project had substantial internal company support, and a very strong business case. The fallout from that executive decision resulted in the loss of a major customer, the loss of several highly talented employees, and several years of subsequent squabbling within the company as it attempted to deal with the business case, which had simply failed to go away. And all this was done in the name of ‘the perception of control’.

Clearly the executives taking the decisions did not feel like they had control over that specific project. And that was in fact quite true. But to achieve the nirvana of stable management for the benefit of the company, true delegation is a necessity, complete control an impossibility, and the ‘perception of control’ a destructive middle ground.

Mar 3, 2006

SkypeOut versus JabPhone quality

I've just had the opportunity to compare voice quality with back-to-back calls on GSM, JabPhone and SkypeOut. JabPhone wins easily, but there are provisos.

First, the test:

  • Calling from Sweden to a fixed line in Johannesburg, South Africa
  • Three calls, GSM, JabPhone and SkypeOut
  • The latter two on a well spec’d machine connected to a 2Mbit frame relay (not loaded)

The results:

  • Best voice quality was certainly JabPhone, according to both parties (and I asked the receiver before expressing my potentially biased opinion :-)
  • The GSM call was worst in basic voice quality, but did not suffer from the same time-lag and occasional moments of silence that both VoIP services had.
  • Best price was certainly SkypeOut (who probably have a POP in Johannesburg, which I suspect JabPhone doesn't). Skype was about 30% the JabPhone price, which in turn was much cheaper than GSM.
  • JabPhone killed the call with no warning when my credit ran out, which was a bit unprofessional (as are many aspects of the JabPhone service).
  • Overall, all three were passable as voice calls, but none were great.

Because JabPhone has the best basic voice quality, I rank it best overall for longer term potential, since the remaining problems are all things that can presumably be addressed. The service is currently advertised as version 0.2, so I expect it will be much better by version 1.0.

Of course, it goes without saying that these opinions are entirely subjective. I’ve always found Skype voice quality to be a bit weird, even back in the days when I did not have the occasional ‘really-bad’ Skype calls (today’s was a good call). I know of others that disagree and find Skype to be great.

When all is said and done, although I have access to, and use, three different VoIP systems (jabber/GoogleTalk/JabPhone, Skype/SkypeOut and a SIP service with x-lite), I still use Skype the most. So, no matter what I think of the voice quality, the continued convenience and professionalism of the service still keeps them in the winning seat as a service (for now :-).

Next to test: compare some other systems like www.jajah.com and the standard SIP service I mentioned above

Feb 23, 2006

It lives! (finally making calls with JabPhone)


Well, I finally got JabPhone to work. Or rather I should say JabPhone finally started working. It seemed they had some severe teething problems in the beginning, and then worked with Google to make better use of the google federated jabber servers. After that they advertised that everything was back on-line and even gave us early adopters a $1 bonus to our accounts (err..., wow?)

But for me it still did not work. Or rather calls went through once in about five attempts. That was a week ago. Todays test worked first time. Voice quality was good, not as good as pure GoogleTalk, but better than SkypeOut (at least my last few calls with SkypeOut have been rather poor).

My overall impression is that it is heading in the right direction, but has a long way to go. It is not nearly as polished a product as SkypeOut, and having to initiate the call with a text message is a minor inconvenience, but an inconvenience nevertheless. No dialpad is more serious. I view this as an alpha level service. I am a bit surprised it is 'on the market'. Hopefully they get it dramatically polished up before it develops a negative image.

Jan 30, 2006

JabPhone - GoogleTalk 'SkypeOut'

Most people know of Skype and the SkypeOut/SkypeIn services allowing you to phone PSTN numbers and receive calls from PSTN numbers. I, for one, have made quite a lot of use of SkypeOut. When GoogleTalk was released (in beta) last year I tried it out, and felt that the voice quality outclassed skype, but it had no SkypeOut facility, so I was not sure how far it could go. But, being Google, I assumed it would get there soon enough.

Well, it turns out that the Jabber software foundation (which made the jabber protocol Google Talk uses) has created a JabPhone site that is basically a SkypeOut service, with the first 15 minutes free. It started just last week, and has already suffered from problems with overload, but they are getting there. I think the fact that GoogleTalk is based on a free and open standard like Jabber, is an excellent thing, and should easily lead to them providing serious competition to Skype. Skype has a massive lead, but perhaps they will start looking over their shoulders for some new competition from the open-protocol camp.

Amanzi Wiki

Contrary to an implication in an earlier posting, I have been updating my Wiki, mostly with news about VoIP that I find interesting to myself. Take a look at http://wiki.amanzi.com.

Skype/911

I found an interesting blog on VoIP and 911, by Richard Edge. He points out that the FCC ruling has a description of a VoIP provider that clearly includes Skype, implying that the FCC 911 ruling should also apply to Skype, and that the other VoIP vendors might take action against Skype for not complying. However, a more recent blog by Richard also includes the comment that 911 only needs to apply if the service can 'Provide, or enable use of, traditional CPE or CPE that, like traditional CPE, is always on and has dial tone'. Since Skype is clearly a service that is only one when the user wants it to be (switches on their computer and runs the skype client) it might be exempt from 911. I personally think that this is the reason why Skype has never gone the ATA route, and only targets people actually running/using PC's. As soon as they provide a Skype phone that does not need to connect to a PC (ie. runs the client on the phone, and connects with Wifi, for example), they would open themselves to possible 911 hassles (and other regulatory hassles).