IBM's Profit May Rise as Shifting Jobs to India Reduces Costs 1798

Kamal R. Prasad

There's no driver for quality here - just quanbreasty. And the only reason I bring this up is the pattern from the last tech boom in the U.S. - when quanbreasty was again emphasised over quality. That was the main driver for the bust.

FWIW, by "quality", I mean building something which will be useful, not defect density or anything else.

Of course it has no merit at all. But it's not possible to simply relocate into a growing area and expect to reap huge benefeits.

Compebreastion is much less useful that you seem to beleive.

Look, most companies I worked for, less than 10% ( and more typically 5% ) of run rate was engineering of any sort. buttuming that for shrinkwrap outfits this is closer to 20-50%, that is *still* not enough cost difference to make a business case.

Something not worth doing for $100 bucks is *probably* not worth doing for $10. If it's worth doing, it is simply worth doing - to the limit of the thing's ability to generate cash flow.

Doing things for any reason other than active, demonstrated need is how failure happens, no matter how cheap - because time wasted is the time multiplied by 100% of the organizational run rate.

And again, please read the ACM study for numbers of US domestic versus offshore work.

I'd have to agree that CMMI provides only the potential to increase efficiency. And that level 1-2 are far more important than level 5.

We cannot say that without a whole lot more qualification. Code isn't like a widget at Best Buy or WalMart; it works more like a durable good. Code that works is valuable; code that doesn't, isn't, and most customers cannot tell you what they want. This means that the more layers of communications error introduced (by distance, culture, incentives or even language ) will be more likely to reduce the value to the customer.

Price theory only works under certain conditions.

People don't buy that, either. They figure the expensive, overqualified guy will be out the door soon as things brighten.

Erm, no... not really. What happens is that people stop doing useless things - which is the main thing that's good about it.

Doing actual tech work is relatively easy - husbanding a product offering that will make money is much harder - by virtue of complexity.

In econ-speak, you are buttuming subsitutability that's simply not in evidence. IN short, a thing that *doesn't* work is much, much more expensive than a thing that *does* work, because the effort to stop the "production line" to fix it is extraordinarily expensive.

Since all evidence used in predicting error rate is purely anectdotal, we're down to word of mouth and past performance as differentiators.

But it's not simply one-dimensional. It depends.

Econ 101 applies to what it applies to. It doesn't *particularly* apply to software very well.

That's one face of it. The demand, it turned out, wasn't really there. People began building empty startups to flip to the majors, and that just exhausted the available supply of capital.

Sure.

Well, espousing race to the bottom economics *probably* isn't backing a very likely horse to win. Be careful out there :)

-- Les Cargill