Comment by jcrites | Hacker News Reader

Comment by jcrites | Hacker News Reader

Parent Full thread

> Most systems operate in extreme conditions, sometimes bordering on the improbable.

Indeed. Your statement reminds me of an insightful engineering article, "How Complex Systems Fail": http://www.ctlab.org/documents/How%20Complex%20Systems%20Fai... . A few relevant observations from the article that resonate with me:

> 1) Complex systems are intrinsically hazardous systems.

> 4) Complex systems contain changing mixtures of failures latent within them.

> 5) Complex systems run in degraded mode.

> A corollary to the preceding point is that complex systems run as broken systems. The system continues to function because it contains so many redundancies and because people can make it function, despite the presence of many flaws. After accident reviews nearly always note that the system has a history of prior ‘proto-accidents’ that nearly generated catastrophe.

> 17) People continuously create safety.

> Failure free operations are the result of activities of people who work to keep the system within the boundaries of tolerable performance. These activities are, for the most part, part of normal operations and superficially straightforward. But because system operations are never trouble free, human practitioner adaptations to changing conditions actually create safety from moment to moment.

Well, a rocket is a thing that:

1. contains a lot of energy

2. releases a considerable stream of energy at a controlled, steady rate

3. does that in a small system, while flying through the air.

It's a lot easier if you drop one requirement:

1 + 2 = coal-based power plant

1 + 3 = bomb

2 (sort of) + 3 = conventional airplane

All of the above have been done reliably. It's only when you put together all three conditions that you get a rocket, and that's when things get really, really difficult.

The other thing to keep in mind with new complex systems is that they will inevitably go thorough a phase where things are discovered over time.

At the start "n" is small 1 flight, 5 flights etc. It's pretty much a given that there will be failures it's just a matter of when and how critical a failure. The failures would be loaded toward happening early on and then fixed so that one particular failure shouldn't happen again. Something else might of course.

As an aside this is similar to why large companies have really long involved legal contracts. They are large enoug,h and have run into enough problems, (or employ attorneys who have) that they've seen it all and done it all and know what they need to do to protect themselves because of the statistical chance that in their large operation a particular problem will appear.

In other words more likely that with a footprint as large as McDonalds there will be a customer that get's scalded by coffee and files a lawsuit than the local coffee shop on main street USA. More customers means more chances of some outlier even happening.

This makes me think of all those crazy warning labels that warn you not to do some incredibly stupid thing (ie "don't use the hair dryer while sleeping"). You know every dumb warning has some sort of story behind it. :)

My dad used to tell me, "Try and live your life so that you never end up inspiring a warning label." .

Is it cheaper to pay by adding extra complexity into legal contracts, extra labels, and extra training across organization vs just paying up for fixing the consequences of an outlier accident?

The point is beyond an unintuitively small size outliers become practical inevitabilities. This isn't black swan stuff either, but just simple percentages.

If you're a small company you generally get to deal with each and every case, and do risk assessment up front on an individual basis. When you're big you're doing enough jobs to standardise those assessments (getting the economies of scale going), processes, and any resulting contracts. The problems get going when the supporting bureaucracy has lost track with reality and becomes a self perpetuating beast, but at least in the early stages it should be beneficial.

"Inevitable" does not mean "must prevent no matter what cost to prevent is".

"when the supporting bureaucracy has lost track with reality"

That's really the key point actually. Things become self perpetuating without anyone even questioning if they are still even necessary.

That said most people don't read contracts (at least most consumers) so in a sense and in certain situations it pays to stuff as many things in there as possible.

What's interesting is how many risky situations you enter into with potential liability where you don't sign a contract.

For example nobody signs a contract to eat a meal at a restaurant. You don't sign away your right to sue if the food is bad or if the meal isn't delivered on time. The car company that sells you a car doesn't require you to sign a contract with them the dealer of course has a sales contract which is different. Car company is big enough to not really have to go that route. They know they don't need a contract to protect against the little things and with a big thing it doesn't matter.

Otoh an electrician just quoted a job to me for under $800. And he sends me this long involved contract (that no doubt he copied from another electrician) which really isn't necessary for this type of job.

I remember when a developer I hired to do work started many years ago. It was a $60k job. Long and involved contract. Ended up signing it (needed the work had no choice). Well we are still doing business 12 years later. But it evolved into a handshake. Nothing in the contract says he has to take my call at 11:00pm on a Saturday night but last weekend he did just that. He and his team finished the repair by Monday morning. I told him to take his wife and family to dinner and we'd pay the bill. Not required but I was really appreciative of the effort. He thanked me and told me it was his job to make sure things were working but he appreciated very much.

Reply on news.ycombinator.com