Agile will fail

Because none of the business-side people bother to learn, and all the commercial tools allow them to avoid learning.

User stories that are written solely from the person at the keyboard, with no thought to the different parts of the architecture - so all the estimates require input from UI, middleware, and server-side staff, and can't be subdivided by the tool into group-specific work

Business-types who cannot define a feature with any precision - "customer will get a message with the offer", neglecting all the many ways the system needs to not repeat the message.

More Grist for the Mill

I saw this today, pushing for more and better programming education for UK schoolkids.

I'm ambivalent about it - I like that it's for strengthening programming, but I feel that the globalization of software development makes it disingenuous to try and get more programmers in the developed nations when all the corporations are looking to ship the programming jobs out to less-developed nations where developers are cheaper.

The conspiracy theorist in me suspects that the CEOs who keep pushing this without bringing programming jobs back home are just looking to hedge their bets in case India and China become too expensive; having a lot of unemployed programmers at home will make it easy to offer lower pay.

---

The Truth, The Whole Truth, And Nothing But The Truth

It's been a demotivating few months at The Job lately, what with the Google rejection, and sundry other depressing events of the new cow-orker variety.

But today I got a snootful of it.  One senior colleague explained today in a meeting that the feature that I was supposed to pull into my server 2 years ago, that would solve all my performance problems, in fact will not solve them as it is currently written, but requires more work to do what it was promised to do 2 years ago.

In the meantime, every time the performance issue was discussed in conversations where this colleague was around, this feature was touted, and my lack of action was wondered at.  Nevermind that changes to the core handling of a major process is not taken lightly; nor is it attempted when customers are getting bi-weekly update releases that cannot fail to work.  I was an idiot for not pulling this feature into my code immediately.

So the tip for developers is this - never lie to your colleagues about what you code can do for them.  If it is not complete, or plain wrong, shut up about it.

---