Originally Posted by Splinemodel
You don't get the economics.
I do get the economic and I understand that you are trying to argue that because software distribution costs are lower (not zero as you imply) therefore total quality should be higher. The two aren't connected directly.
When you have something with 0 COGS, you can amortize the development dollars easily over large quantities of users. Planes do not have 0 COGS. In addition, planes of course have design-related cost but also manufacture-related cost, and the issues that go with. Software does not have manufacture-related cost or issues.
This is false. Productizing a software item does have costs and manufacturing and distribution costs are not zero. Secondly, development costs of aircraft are pretty high. The 787 has an estimated pure development cost of $16-18B (including all the cost sharing with Boeing's partners...about $10B for Boeing alone).
I've done both in house and external software product development. I can attest that the latter is more expensive than the former and the costs are non-zero.
I would argue that the design of Windows took more dev dollar than the design of, say, the 787.
I would argue that a cursory google would have yielded that Vista cost MS $6B to develop and that $6B is lower than either the $10B or $16B estimates for the Dreamliner.
The estimated would-be dev cost of Linux is $1B, according to sloccount.
Which is bogus. I personally have around 50K sloc tabulated in Ohloh and personal knowledge of the dev costs on two open source projects (where all devs paid by a company) and the Ohloh estimates are at least an order of magnitude wrong. These are similar estimation methods to those used to quantify Linux development value.
To me this screams that software design and development paradigm are inherently flawed.
Bullshit. The value proposition of software is as high, if not higher, than comparable disciplines. think I quoted references three years ago so wont again.
I blame C/C++/C#/Java. Stupid-shit. Using the C-family of languages is like building a 787 with hand-tools. Studies repeatedly show that languages like Ada shave off massive amounts of software dev dollar (and it runs hella faster than Java). So a it costs less to design a super-reliable system. It's quite apropos, given that most flight control systems are still written in Ada.
I blame your superficial understanding of software development to make such broad unfounded assertions. Ada does not shave off massive amounts of software dev dollars, if it did, then industry would have adopted it. Instead, even DoD has moved away from it except for some real time systems. It has advantages in this specific domain but the software dev costs for this domain are significantly higher than normal software development.
I've done flight system development. Have you?
So that's the 180. This thread is years old. I'm not knocking SW developers, just the paradigm. I believe the paradigm is inextricable from the language.
And you would be incorrect in your belief given that these are all still procedural languages you're talking about and there's no paradigm difference between them from the perspective of development methodology.