You’re reading Signal vs. Noise, a publication about the web by 37signals since 1999.

Wrong. Look no further than biology.

Evolution goes BOTH ways.

I think it’s probably referring to software… ;o)

Nah, he’s referring to lots of things “Systematics”, as he terms them.

But there’s certainly good counter examples in industry/management, and probably in software projects too.

Nice. I do agree. Although, we do have a CRM here that has evolved into a monster (maybe it was always a monster, just smaller).

Starting from scratch is the better option, but try telling management.

Interestingly, even in biology, isn’t it the theory that life started simply – single cells, for example – and evolved into more complex structures only after the simple systems proved that they worked (survived in their environment)?

I’ve been looking for this quote for years, I had attributed it to Martin Fowler, but now upon second thought I might have read it in book by Martin Fowler.

There is no other way to do it. I’ve tried. Start simple or FAIL !

Nonsense. The world is full of horrifically complex accomplishments that started out that way. They’re often not perfect, but they sometimes do “work”.

Sometimes there is just to big a gap for simple to work at all.

The world is full of horrifically complex accomplishments that started out that way.

Yes, the world is full of accomplishments that are horrifically complex. But each is a part of a system that has been refined and expanded over time.

By definition, a complex system is made up of multiple parts… so yes, it probably goes without saying that the simpler parts need to be designed first to create a complex system. I can’t think of an example of a complex system built from scratch.

Jebus. These ideas are dated, people. If you’re interested in this stuff, be interested in it.

An arch is irreducibly complex. That is, you can’t build an arch without scaffolding, but once the arch is built, the scaffolding can be removed. Thus a simpler, working system has evolved from a more complex one.

There. Now move on.

In other words, so-called “irreducible complexity” doesn’t exist. As an evolutionary geneticist, I’m inclined to agree, with the proviso that complexity may emerge with the aid of scaffolding that is later stripped away, leaving the origin of the complexity mysterious to subsequent observers.

the whole idea seems rather ambiguous.

you can argue that a single-celled organism is very simple, yet the forces that sustain it are quite complex are they not?

but assuming we’re talking about software or similar, just because something generally gets more complex over time, why are we assuming the original system was “simple?”

Ah, “Systemantics” (now known as “The Systems Bible”). One of the best texts, ever. I still owe a debt to the boss who introduced me to it over 20 years ago.

And never forget, “Fail safe systems fail by failing to fail safe.”

And “All systems have intra-system goals.”

And especially “Intra-system goals come first.”

@Tim Case: It seems Grady Booch likes using that quote in several of his books.

I think the most obvious example of something starting out with complexity, and working correctly, was Creation (of the world, man and woman, animals, etc). Believing single cells miraculously became more complex gradually over time belies a serious lack of understanding the complexity found in a single cell in the human body. So many of the micro-systems of the body require all (or most) parts to be present to work correctly.

But chances are, where man is involved, your quote is accurate.

a single celled organism isn’t simple at all. The protein folding encoded by the DNA alone is more than a world full of supercomputers has yet been able to crack. The RNA and other, uh… “nanotech” support systems are incredibly complex, not to mention the various energy transfer systems.

I find the quote inaccurate for software systems too. Sure my psuedocode may look simple, sure I may build from the bottom up, but not really – I’m several layers of abstraction from what is really going on with the electrons in my CPU , not to mention the various support systems that keep my machine/network/internet running.

I do agree. I’m inclined to agree, with the proviso that complexity may emerge with the aid of scaffolding that is later stripped away, leaving the origin of the complexity mysterious to subsequent observers. I can’t think of an example of a complex system built from scratch. Its a part of a system that has been refined and expanded over time…..!

If we interpret “a complex system designed from scratch” as “a complex system by design” it makes a lot of sense.

A system is hopefully designed before it is put into effect. One that is complex in the design phase, before reaching production even, is likely doomed to fail.

A good example of this is our credit system.

A good example of this is our credit system.

Really? I mean, really?

How so?

“Fail safe systems fail by failing to fail safe.”

Here’s a good literary example.

I assume this is a quote in his book. He constructs this complex mess of words when all he means is “Fail safe systems crash hard.”

BUT , did he construct the final sentence from the simpler idea? Point is…it’s impossible to prove this as right or wrong, in software or in nature. However, EVERYTHING in our universe is made up of smaller parts.

to accord with either the relative or absolute is still not enlightenment.

Biology is a bad example. I don’t know of a single scientist who could build a single human, animal or plant cell from scratch let alone a whole living organism.

As far as software is concerned, I wish I could tattoo this on the back of my developers eyelids. Every project that I have seen where people were trying to migrate from an old VB6 application to a .Net application, in one big bang , with all of it’s functionality has failed or come in 4 years late(Which is still a failed project. This usually comes from smart ass developers who think they can do it but invariably fail due to lack of requirements or poor understanding how the old application does things without the proper documentation. The successful projects all start from scratch and release something small at first and build up over 2-4 years. Just my experience working my way up from developer to CTO , but what would I know?

Who is John Gall?