Reply to: Interesting Read From the Guardian

[Mich the Tester]

One thing I have seen quite a few times was when mainframe organisations moved to the web in the late 90s and let the architects build core systems based on, it would seem, a quick read through, "java in 24 hours" and a few javaworld articles. A lot of these systems are still kicking about, central to the business and absolute monstrosities to work with.

I know of one place that has a a 283 column database table that holds pretty much eveything and if you have went through Uni in the last 20 years you will probably be on it. Of course performance is terrible so they came up with a plan to add another column of nulls and they believed that this would free up memory and thus increase performace. These people should be removed from their jobs.

No they shouldn't! They provide work for contractors!

You're right, they should be removed from their jobs, and in fact kept away from computers, the internet, fragile manufactured products, road, sea or air traffic, perishable food goods or anything else where they can damage other people.

Unemployment benefit wasn't invented for them; it was invented to protect the rest of us against them

[minestrone]

One thing I have seen quite a few times was when mainframe organisations moved to the web in the late 90s and let the architects build core systems based on, it would seem, a quick read through, "java in 24 hours" and a few javaworld articles. A lot of these systems are still kicking about, central to the business and absolute monstrosities to work with.

I know of one place that has a a 283 column database table that holds pretty much eveything and if you have went through Uni in the last 20 years you will probably be on it. Of course performance is terrible so they came up with a plan to add another column of nulls and they believed that this would free up memory and thus increase performace. These people should be removed from their jobs.

[alluvial]

The prediction that systems will get too complex and will cause a crisis have been going since the first wave of computer development in the 1960's. It's like peak oil and climate change, at whatever point you look at it people say the same things "we're too dependent, it's too complex and there will be a crisis within the next 10 years". That was the original reason Computer Science became a key subject area at Universities, in order to tackle the crisis. Of course we all know how influential Universities are in computer and software development. How many project managers go running off to a University to get some help from a professor of Computer Science?

I don't believe that there will be a single crisis that brings down all the world's computer systems. I do believe though that we will see more occurences of major organisations having incidents that cause anything from serious embarassment to commercial diasater.

[Mich the Tester]

The prediction that systems will get too complex and will cause a crisis have been going since the first wave of computer development in the 1960's.

Arguably they already have done. While I wouldn't suggest the banking crisis or the war in Iraq were caused by computers, which would be ridiculous, an unfounded belief in the correctness of algorithms certainly played a part, as explained in this documentary series.

Anyway, a good systems failure is never a bad thing for contract testers (as long as you're not on that project before things fail).

[alluvial]

Nice in theory but costs too much. There are only a certain number of years someone will put up with 0-2% pay rise.

Is that not a contradiction?

Which is why the old way of doing things was better. There were teams of developers putting in new systems and they supported and resolved any issues with them. The knowledge resided in the team and was shared between them. Even if people moved on, there was a sort of dynamic equilibrium as new people came into the team, learnt the ropes and took over when people left. Things started to get screwed when the development and support functions got separated.

[BlasterBates]

The prediction that systems will get too complex and will cause a crisis have been going since the first wave of computer development in the 1960's. It's like peak oil and climate change, at whatever point you look at it people say the same things "we're too dependent, it's too complex and there will be a crisis within the next 10 years". That was the original reason Computer Science became a key subject area at Universities, in order to tackle the crisis. Of course we all know how influential Universities are in computer and software development. How many project managers go running off to a University to get some help from a professor of Computer Science?

[minestrone]

Nice in theory but costs too much. There are only a certain number of years someone will put up with 0-2% pay rise.

Is that not a contradiction?

[SueEllen]

I strongly believe that the designer of a system should be onboard from inception to EOL so as to maintain core architecture princicples and to see the errors of the initial design for their own learning. If an architect or a developer never has to deal with the problems they create at start then they never see what they produce as problamatic.

Nice in theory but costs too much. There are only a certain number of years someone will put up with 0-2% pay rise.

[minestrone]

I strongly believe that the designer of a system should be onboard from inception to EOL so as to maintain core architecture princicples and to see the errors of the initial design for their own learning. If an architect or a developer never has to deal with the problems they create at start then they never see what they produce as problamatic.

Ach, who cares, invoice day.

[doodab]

I think the trick is to ensure you are the manager who is there at the start and middle of the project, and moved on long before it's delivered.

There are way too many of those around

[SueEllen]

That's interesting that. I do wonder just how long it can go on before the idiots in charge finally realise that they are spending money hand over fist for their development work. Not just the initial development but onshore fix costs and higher maintenance costs.

I think the trick is to ensure you are the manager who is there at the start and middle of the project, and moved on long before it's delivered.

[alluvial]

Current client co have learned their lesson with that, seen a £100,000 estimate for a system to be built in India hit £10 million and when it got back onshore it cost a good few million to fix. The internal team quoted 8 million for build and management laughed and signed up with the bobs.

Client co no longer gives first version work to India.

That's interesting that. I do wonder just how long it can go on before the idiots in charge finally realise that they are spending money hand over fist for their development work. Not just the initial development but onshore fix costs and higher maintenance costs. I have heard of a couple of other financial institutions that are bringing work back onshore and hoefully it will continue.

Who was it that said if you lose control of your IT, you lose control of your business?

[minestrone]

The problem is that we offshore the development as well as the maintenance which only makes matters worse.

Current client co have learned their lesson with that, seen a £100,000 estimate for a system to be built in India hit £10 million and when it got back onshore it cost a good few million to fix. The internal team quoted 8 million for build and management laughed and signed up with the bobs.

Client co no longer gives first version work to India.

[doodab]

The Basic Laws of Human Stupidity

POTD.

[Bunk]

I see the same problem with software, Britain is in no way capable of maintaining the software it requires to function by internal means, our systems are so badly written and so unmaintainable that we have to offshore and I see that trend exponentially rising.

The problem is that we offshore the development as well as the maintenance which only makes matters worse.