Sunday, October 21, 2012

'The Art of Persuasion', Bugs, and Power

Someone once said to me that I should read, 'The Art of Persuasion'. It's supposedly a classic and shows people how to influence and shape conversations/organisations for mutual interests. In reality it's not that simple though. Many people and organisations often have vested interests making decision making extremely complex and difficult especially when there are potential veto powers involved. This problem is exacerbated by cultural issues which makes discussions between managers and sub-orindates difficult if not impossible even if the technical technical arguments are significantly in favour of the sub-ordinate (Personal experience has indicated that a lot of the time by the time you've convinced them it's too late or you've probably moved on to other problems by then in which case you're only choice may be to simply get on with life.). Moreover, depending on the nature of the argument and cultures involved compromise is not only difficult but sometimes totally unacceptable. Take the territorial dispute between China, Japan, Taiwan and others as a example.

Lets take a more pragmatic/practical perspective. It's land (you could argue that some of the islands are merely large rocks) that can be divided/shared but if you were more cynical you could maintain the status quo (leave discussion about the land for a later time) and merely divide/share the proceeds of any mineral resources based on further discussion and once mining were underway. The way I see it the change in leadership in some of these countries is an opportunity, not a hindrance. They and many other 'Lame Duck' heads of state are doing themselves and their countries massive a massive dis-service. They have an incredible opportunity (provided they have the requisite powers and support) to be able to do something purely for the benefit/sake of their country without having to deal with the potential awkwardness of a successive term.

While we're on 'The Art of Persuasion', one of the things that was a little bit surprising to me was the relative difference in the standard of some graduates in IT. Some of them are quite good, others not so. However, with recruitment sometimes not being as rigorous (short tests, interviews, and referee checks can all be circumvented with the right preparation) as it should be it sometimes is hard to find the difference between those who can pass the test and those who can truly excel given the right opportunity. So I say use proper projects as a means of guaging opportunity. A two week University project is nothing in comparison with a multi-month/year long project. Pick something, anything (game, search engine, trading system, operating system, etc...). Don't train the prospective candidate but pay them enough money for them to survive and then let them build it. Maintain regular reports to reduce the chances of plagiarism. It would amount to a apprenticeship/postgraduate education without many of the risks associated with them (time/money). If they're good enough give them an employment offer. If not, at least they have a project they can take away to show to another employer (make the project shorter to reduce the chances of exploitation).

If you think this is unfair or overly difficult then let's put this into perspective. In my recent history I've completed about 1200+ pages/300K+ words (6 thesis/dissertation length research projects) of private research work (with progress records) so I don't have any doubt that any reasonably skilled, motivated graduate should be able to do something similar.

You've probably discovered that I regularly discover software flaws. Here are my 'Bugs of the Week'.

I sometimes use a Direct Attached Storage (DAS)(Thecus) device. However, I also use a require a NEC Electronics USB 3.0 Host Controller ExpressCard to use it to it's maximum capibility. The DAS uses a particular driver which is also used by another piece of background software which seems to manage/monitor drive caching to boost efficiency efficiency. Several times I've tried removing the device but sometimes the monitoring software doesn't want to unload the driver from memory even though it is doing nothing with the device. If the behaviour was predictable it would be fine but in this case reverse engineering is required to figure it out. Removing it/replacing it with an alternative has proven to be a simpler option.

MediaCoder doesn't play nicely with Windows 7 Firewall Control (finer grained control of outbound behaviour of Windows network behaviour in particular). If you don't create an enabling rule you can't open up the Setting page from the 'File' menu.

Known about this one for a while now. Older version of Opera on Debian platform isn't able to deal with having ~/.opera folder deleted while program is still running in memory elegantly. Program crashes without enough detail to debug. Problem has been fixed in subsequent revisions.

Not so much of a bug but still annoying. When programs run out of memory on Knoppix they simply crash quietly without warning. 'dmesg' gives you the details but surely there should be a service/daemon which monitors memory usage and gives you prior warning before a crash?

Adobe Flash plugin under Opera/Knoppix doensn't play nicely with sometimes. Under certain circumstances (during playback of certain video files) it will simply send you to another TTY (Alt + Ctl + F? to send you to the right one). At first thought they were terminal kernel panics/oops. Examination of message said otherwise. Will need more time/examples to determine exactly what is happening.

Surprising how much electricity some appliances use.

Continuing work on 'Cloud and Internet Security' report... Will post back further findings here at another time as the amount of content is stacking up and summarising can sometimes be time consuming.