Fog Creek Software
Discussion Board




This is why software will continue to suck.


From a Computerworld article dealing with offshoring companies offering consulting services:

"  And a decision on what vendor to use "comes down to cost," said Don Weiner, managing director and global head of technology at Deutsche Bank in New York.  "

Comes down to cost...Not cost, experience and quality. Nope. Just cust. The cheapest SOB wins the deal.

So, if you are in the employ of Deutsche Bank in New York the message is loud and clear: We really don't give a shit about quality. That's a problem for the next guy. Today all we care about is cheap. Cheap. Cheap. We want to look like saviors at the annual stockholder's meeting so we're gonna cut cost to the bone. Quality isn't important. Cost is.

Nice.

Beam me up.
Thursday, June 24, 2004

Along the same note, remember: our military is equipped by the lowest bidder.

BeamMe2
Thursday, June 24, 2004

Yes, but the military has very strict standards on quality. I've worked for a company that supplies stuff the military. Very strict, precise requirements.

Not so with software. Vague requirements and easy to hide bugs. All at the lowest cost.

Beam me up.
Thursday, June 24, 2004

No, the military is NOT supplied by the lowest bidder.  The Military is allowed to consider the likely-hood of delivering, and earlier contracts the vendor has delivered on.

And, as has been pointed out, the Military has a rather high quality bar (usually they do, Strategic Missile Defense excluded).

AllanL5
Thursday, June 24, 2004

...the lowest bidder that can meet the conditions of the bid request.

"Provide 1,000,000 hammers weighing between 9.5 and 10 oz with a head diameter of 22.7 mm and a shaft length of 35 cm, which can withstand rapid temperature fluctuations between -30F and 130F at a delta rate of 60F per hour. The hammer must be non-metallic, and much withstand a high intensity test of 1,000,000 44lb blows with no material faults."

Paying more just for the hell of it doesn't get you a better product. The key is carefully stipulating exactly what you want such that it is impossible for a vendor, expensive or cheap, to deliver subpar products. 

Dennis Forbes
Thursday, June 24, 2004

Good software is VERY expensive.  The software for the space shuttle cost $1,000 per line. 

Myron A. Semack
Thursday, June 24, 2004

"Comes down to cost...Not cost, experience and quality. Nope. Just cust. The cheapest SOB wins the deal."

Generally speaking, no one gives a damn about quality. Cost is the only concern, period.

Why should the sw industry be different from the rest?

No, I don't think this is the way to go. Yes, I ocasionally pay more for a product, by not buying it at the super-discount mega-store, because I prefer the service I get at my local "Mom & Pop".

"Mom & Pop" doesn't automatically equal higher quality, naturally.

Paulo Caetano
Thursday, June 24, 2004


"Generally speaking, no one gives a damn about quality. Cost is the only concern, period."

Hopefully, the architects building our bridges and skyscrapers don't subscribe to that notion.

I don't think anyone is advocating a pointless pursuit of the highest possible quality in software. Rather, what I'm saying is that while people bemoan the lack of quality software all the while it's becoming more pervasive in our lives and businesses. So it makes sense to say "Hey! Maybe quality in software does matter, along the same lines as quality in building a bridge matters."

Beam me up.
Thursday, June 24, 2004

Dennis, i don't see a handle diameter for the handle.
So i could make the handle large enough that nobody
could hold it as long is met the other specs?

That's the problem with a requirements/black/white/cost
approach to life. It's hard to say stuff like and oh,
use common sense appropriate to the problem; have
a good corporate structure and people that will make it
so you can actually get the product done, support it,
extend it, and be nice for the cutomer to work with.

son of parnas
Thursday, June 24, 2004

The problem with commoditizing software by definition is that software is impossible to define rigorously up front as you can a piece of hardware or any other physical object. There are no universally accepted AND understood methods of defining a piece of software. Quality of software is likewise intangible.

But both the characteristics and the quality of software are absolutely critical. If the functionality or the quality are not there, then the effort may be entirely a waste to a much higher degree than any acquisition of a physical good. IE: there are surplus markets for unacceptable physical goods. There are no surplus markets for crappy software - nobody will pay for it.

That is the underlying problem of commoditization of software: to someone who doesn't know any better (IE, executives out of touch with day to day operations who make outsourcing decisions) it becomes very easy to cheapen what you (they) can't see, feel, quantify, nor understand.

Bored Bystander
Thursday, June 24, 2004

son,

I was commenting only in regards to the statement that the military equips soldiers with the lowest bidder equipment. The premise, of course, is that the equipment is subpar, when in reality what the "lowest" bidder is bidding on is very precisely specified (and quality is one of the specifications). Someone else mentioned civil infrastructure, and the reality is that bridges, roads, sewers, etc, are often handled in the same way - specific requirements are laid out, and the lowest _credible_ bidder wins.

Contrastingly, it is next to impossible to create specific software specifications, just as it is next to impossible to create specific specifications for fighter jets. In the latter case the most credible bidders are sponsored for a prototype, and then they can prove their ability to deliver, and the special advantages to their approach, and this is weight against the price.

Dennis Forbes
Thursday, June 24, 2004

Guess what OSS is low cost and generally of good quality.  It can be done.

Software Assurance
Thursday, June 24, 2004

>>Guess what OSS is low cost and generally of good quality.  It can be done.

In the instance of Linux, the prestige of being associated with core code is a form of non monetary compensation that developers covet.  In fact the highest profile OSS projects - Linux, Apache, mail agents - have very high documentation standards.

Lower profile OSS projects are often crap and have spotty to nonexistent documentation. Why - no prestige factor.  Nobody notices if the job is well done or not.

And internal IT type commercial projects have exactly zero prestige factor. So the client pays for every line of code.

Bored Bystander
Thursday, June 24, 2004

The best programmer is 28x as effective as the "worst" programmer. (per a recent SW book, who's name I forget).


And since the best programmer doesn't cost 28x a much, they're a good deal.

Only solution is for those who KNOW this fact (perhaps only programmers?) to hire those good people.

Or for those excellent programmers to start thier own companies.

But guess what? I bet the best programmers, who start thier own companies, don't realize that the best marketing and sales people are 100x as effective as the least.

I know one very skilled programmer who REFUSES to advertise his software.  And if some competition enteres his market, he simply moves to a different product/market.

Stupidity abounds.

Mr. Analogy
Thursday, June 24, 2004

<>Lower profile OSS projects are often crap and have spotty to nonexistent documentation. Why - no prestige factor.  Nobody notices if the job is well done or not.<>

You could say the same thing for windows shareware/freeware.

<>Hopefully, the architects building our bridges and skyscrapers don't subscribe to that notion.<>

Well indirectly they are thinking of the cost: people sueing, lost businesses hours, repairs they will be forced to do, etc..

somemorone
Thursday, June 24, 2004

I think people understand that great software is expensive. What they don't understand is why, and how it affects the cost in the long run.  Good software costs a lot because it's GOOD.  Cheap software is bad because your requirements probably suck, so someone can create something that looks like it's doing that for practically nothing.  If companies came out with good enough specs on software, the lowest bidder would be WAY higher.  Maybe people need to get a lot better at understanding how to write specs.  As it is, pretty much only great developers could write great specs, and if you already hired great developers, you wouldn't need to contract out the development, right?

Devin
Thursday, June 24, 2004

Mr. Analogy

May I use your post for citation?

René Nyffenegger
Thursday, June 24, 2004

"There are no universally accepted AND understood methods of defining a piece of software."

That's because software is nothing more than an extremely precise specification of a solution to a problem.  Nothing more, nothing less.

Jim Rankin
Thursday, June 24, 2004

>> That's because software is nothing more than an extremely precise specification of a solution to a problem.  Nothing more, nothing less.

Jim, that is VERY quotable and oh so true. This is one of those statements that I read in which initially I think "nahhh, shallow BS". No... the software IS the specification of the solution.

And it's pretty well impossible to consistently abstract out a much simpler specification of the problem solution than that which comprises the actual code.

Bored Bystander
Thursday, June 24, 2004

I for one think its great that banks go with the lowest bidder. I work for a consortium of international criminals. We have redesigned transaction systems for a dozen international banks so far -- we subsidize the development cost with our own funds, hire top notch developers, and deliver excellent results that please and thrill the customer. It also works well for us as we have many untraceable systems for money laundering as well.

William the Kid
Thursday, June 24, 2004

"We really don't give a shit about quality. That's a problem for the next guy. Today all we care about is cheap. Cheap. Cheap. We want to look like saviors at the annual stockholder's meeting so we're gonna cut cost to the bone. Quality isn't important. Cost is."

Sounds like my clients.  ;)

Norrick
Thursday, June 24, 2004

One hopes that as a financial institution, they mean "cost over the lifetime of the product".

If a banker buys a car, and the car salesman offers to use super-cheap tires, will he get those?  I hope not.

Similarly, if presented with a program of lower up-front-cost, I hope they'll apply the same logic they do elsewhere, and not assume it has a lower cost, if, say, it's riddled with bugs that will make employees spend twice as long doing the same task.

kyle
Tuesday, June 29, 2004

As the originator of the quote at the ITAA, you guys bash for bashing sake.  You took the quote totally out of context, just like the author of the article did.  All things being equal (skill, experience, attrition rates, etc.), why NOT choose the lowest cost provider.

Quality is #1 to all tech execs.  However, since our budgets can't afford perfectly error-trapped code, and straight-line code is too risky, there needs to be a happy median between functionality, cost and quality.  Not easy to do, but as a tech exec wanting to minimize maintenance costs, quality is important.

You should also be aware that the article continues with saying the subject matter expertise, alon with architecture and design are kept inhouse, and not outsourced.  Its the commoditized software coding that goes.

Me
Tuesday, July 27, 2004

*  Recent Topics

*  Fog Creek Home