patrick.net

Proof that if you give a million monkeys a million keyboards, they will NOT eventually produce anything at all of value, let alone the works of Shakespeare.

Dan8267 says

25 to 50 million lines of code? It would not take even 1 million if done right. A million lines of code is a hell of a lot in today's programming platforms.

Not if every class has to be inherited from a bloated base class, that uses third party components and has 5 to 10 List collections, and also every class implements 4 or 5 Interfaces.
The IT shops today can blow right past that 50 millions lines of code, and god forbid you write a lambda expression to accomplish a task rather than writing a new Interface and a class to implement that interface, with virtual methods then override it in a third class. why you'll just blow the Junior programmer's mind and the CTO wont be able to follow you're code review.

CaptainShuddup says

and god forbid you write a lambda expression to accomplish a task rather than writing a new Interface and a class to implement that interface, with virtual methods then override it in a third class. why you'll just blow the Junior programmer's mind and the CTO wont be able to follow you're code review.

Lambda expressions have become more popular again with the renaissance of functional programming, so there's hope ;)

Big picture people:

Software development is something that 99% of the human population couldn't do to save their lives. Out the remaining 1%, 99% cannot do software development at a level acceptable for commercial use. Only about 700,000 people in the world can do software development at a commercially acceptable level. And only about a tenth of them, 70,000, are truly good at it.

And that's why good software developers are worth much more than their weight in gold.

*Statistics derived from empirical observation.

Dan8267 says

And that's why good software developers are worth much more than their weight in gold.

Unfortunately, they don't work for Microsoft.*

*obvious to everyone

This software project has the look and feel of a minority contractor.

Dan8267 says

Big picture people:

Software development is something that 99% of the human population couldn't do to save their lives. Out the remaining 1%, 99% cannot do software development at a level acceptable for commercial use. Only about 700,000 people in the world can do software development at a commercially acceptable level. And only about a tenth of them, 70,000, are truly good at it.

And that's why good software developers are worth much more than their weight in gold.

*Statistics derived from empirical observation.

Thank you.

Dan8267 says

Out the remaining 1%, 99% cannot do software development at a level acceptable for commercial use. Only about 700,000 people in the world can do software development at a commercially acceptable level

Dan, do you not believe that our scientists and engineers, many of whom, underemployed as lab assistants, postdocs, adjunct profs, paper pushers at Dow/DuPont, etc, can become good software developers?

Or is there some special talent, aside of dissecting and solving technical problems which scientists & engineers have been doing for decades?

Dan8267 says

And only about a tenth of them, 70,000, are truly good at it.

And that's why good software developers are worth much more than their weight in gold.

Not sure about the statistics but I find software development is generally easy to do. Many are doing it as kids. Big US companies rely massively on minimally trained engineers from populous countries to do most of their developments - granted with mediocre results software-wise, but good enough money-wise. Generally all it takes is a good architect to organize things upfront and a lot of QA testing to eliminates loads of bugs. Unfortunately the software professions have become commoditized in that way. Most software products nowadays are poorly coded but generally get the job done regardless.

The argument "there are only so many of these guys" applies to pretty much any profession. You could say there are only so many good bakers therefore they are worth their weight in gold. Well maybe they are, bakers.

Heraclitusstudent says

Generally all it takes is a good architect to organize things upfront and a lot of QA testing to eliminates loads of bugs. Unfortunately the software professions have become commoditized in that way. Most software products nowadays are poorly coded but generally get the job done regardless.

Well, the architect piece is why I'd brought up the scientists/engineers analog.

In my former field of study, applied chemistry/chemical engineering, we're trained to think in terms of mass & energy balances and then, degrees of freedom within designing a chemical process. And thus, there's a method to the madness which has theory, "algorithms", and performance analysis, throughout the process, not just in the QA part.

This is why I don't understand why we don't have good programmers when many engineering majors are trained to think in terms of problem solving and analysis.

The two best software engineers I have worked with, in the sense of the most valuable to the employer, were similar in a lot of ways. She was pleasant, helpful, and a strong proponent of sound architectural practices which saved the company a lot of money in the long term. He was unpleasant, bitter, vicious, yet a good influence on the architecture to the same degree she was.

From spec to ship, both were about four times faster than what I'd consider to be a minimally competent coder. And their code usually worked the first time, with only a few minutes to fix the bugs found by QA, assuming it was these two I'm talking about doing the fixing.

Yet both wrote absolutely unmaintainable code. She wrote unreadable spaghetti which gave maintenance coders fits. He didn't know how to indent or choose comprehensible names, and was profoundly unsound at dividing labor among modules.

(Curiously, although everyone was ready to strangle him, his larger sense of architecture was so solid that it was difficult to deny his brilliance.)

So both were good in the larger sense, and great if you never need to change or maintain your code, but fairly poor at the minutiae. Without a few patient and intelligent maintenance coders, their employers would have been in trouble.

I don't believe there is such a thing as a software methodology. You need generally intelligent engineers with high cognitive ability, low self-esteem, and an ability to communicate clearly.

As for the existence of super-coders? Don't know if they exist, and, since nobody can write any reasonably-complex commercial application on their own these, days, I don't know what good they'd do any firm. A good engineer must be willing to explain clearly, to listen, and have decent social skills.

By the way: usually when I hear management singing the praises of some super-productive coder, I find that they ran with an ill-informed idea, carried it to perfection, and then quit or moved to another project to leave the now-flailing project to others. Things like massive database applications written entirely in Perl, with everything processed one row at a time. And the fastest programmers prioritize speed over quality to a high degree.

Go look at the autograph of *The Marriage of Figaro* if you ever get a chance. Six weeks' work. There are likely Mozarts in the software world, but they're rare.

zzyzzx says

This software project has the look and feel of a minority contractor.

It was the usual collection of creaky, bloated, and well-connected firms the federal government is compelled to hire simply because every congressman has always gotten checks from them. And these companies are known to be inefficient and incompetent from their past cost overruns and failures.

I have read somewhere that SAIC was involved - almost too scary to contemplate. Anyone know?

(But I'm sure these firms hired tons of blacks, illegal aliens, Koreans, and other criminorities because the Kenyan did not want good Christians getting money for this boondoggle.)