Craftsmanship? Design?

The other day's rambling about the built-in seating at Louis' Lunch got me to thinking about the phrase, "They just don't build 'em like they used to."

In meatspace, I can understand how this phrase came about. It's easy enough to look out at the world and see plenty of objects out there that still exist where, if they were built today, the odds would be against them at that age. My house was built in 1871, but still holds up pretty well. I'm not sure the house I used to live in, built in 2004, will still be standing in 2145. It's even more prevalent overseas; there are roads and aqueducts that date back to the Roman era-- but yet, the new bus station in Montgomery County has an expected life of a mere fifty years. Assuming they actually fix the concrete they poured, but that's beside the point.

Outbuildings at Dover Castle have held up reasonably well.

When it comes to actual gadgetry, the "automata" I reference in the blog title, it feels the other way around. It feels like they now build 'em so much better than they used to. I recall in the mid 1980s that it was highly recommended that we quit playing with our Intellivision when the chassis got warm; the electronics inside would get the outside warm to the touch in about an hour. But today,  would't think twice about firing up a modern console and burning the entire afternoon. Well, I suppose I would, but it would be because I never seem to have a whole afternoon to play these days. Back in those 1980s, many people wouldn't buy power windows on a car simply because it was "one more thing that can go wrong"-- but these days, it's hard to find a car that doesn't have them. The hardware's gotten robust enough that it doesn't break often enough that people care. A vacuum tube had a mean time between failures of 5,000 hours. Now we have integrated circuits that have failure rates 100 times better than that.

And strangely enough, it's been for entirely opposite reasons. In the early days of building stuff, there was less engineering knowledge. Certainly, there existed architects, but design was nowhere near an exact science and was likely to change even for reasons that had nothing to do with the logistics of building-- the dome on St. Peter's Basilica, for example, went through different designs because as they got around to building the place to where the dome could go on, the architects in question-- and the popes who hired them-- died. But even looking past that, we find that piers that hold up the dome are bigger than the designs called for. The dome in Michelangelo's final plans was ovoid, and was supposed to be taller than the hemispherical dome we see today. Somewhere along the way, the craftsmen who actually built the thing made some adjustments. The basilica was "built to last"-- we never seem to say "designed to last." 

Now, admittedly, looking at the world, there's also a selection bias at hand. We have thousand year old buildings standing and we look at them and think, "Wow, they must've built 'em really well back then." We think the same thing when we see a fifty-year-old classic car on the street. But all we truly know is they built _that one_ really well back then. For every eleventh century castle that still stands, there's probably a dozen that have had their component stones turned into a succession of other structures over the intervening decades.

Designing toward an intended service life is what seems to have changed most everyday objects and is what's leading us to say that they don't make 'em like they used to. We don't build houses out of 3x7" timbers, we use cheap 2x4s. We shave material off a design to make it cheaper to make (or ship!)  even though it might mean the end user will have to replace the part at some point. Engineering and design have given us the ability to design toward goals other than robustness.

Gadgetry, on the other hand, exists in large part because of that engineering and design. One simply can't hand-build a memory device with billions and billions of transistors that makes postage stamps look brobdingnagian by comparison. Design work towards adding features, increasing speed, and reducing power draw are all welcomed.  We don't have to worry so much about the robustness of a design, though, because it'll be obsolete before the year is out.

Software's wound up in an interesting space where we put so much effort into a design and a process, but our implementation winds up not necessarily being top-notch; it's rushed, and the product gets out the door riddled with bugs. It's not that there aren't excellent crafters of code out there; there are. But we've stepped so far away from valuing robustness that other concerns matter more, like being first to market. Perhaps part of it is that we feel we can fix the problems after we ship, as well.

Once upon a time, software was hand-crafted as well, and it was, legend has it, notably more robust. It seems that in fifty or so years, the software industry has matched up with five hundred years of the transition from craftsmanship to design in meatspace.

And now to the actual point of all this rambling: I wonder if there's ever going to be much point in going back? Certainly, there's no cramming millennia of  engineering advances back into the genie lamp, but is there ever a point where robustness becomes a more important design goal again? Is there a point to designing an object with what's effectively an "unlimited" service life?  Given our knowledge of things, could we even do so?  We now know enough about natural disasters that we build things to withstand the "hundred year earthquake" or the "fifty year flood"-- how would we design a structure to handle "any forseeable natural disaster"? Will we ever once again "build 'em the way we used to"?
With the rise of 3-D printing, I can certainly see a world where we no longer head to a local store to buy small objects, we instead buy a design for 'em and print one off at the house. Will that change the way we design things and bring back robustness as a design goal?