The software equivalent of this mistake would be spending millions of dollars building software targeting Linux when the customer needs to run it on Windows. It's not "certain input parameters", it's obvious environmental conditions that should have been considered at the outset.

I would really like one of the naysayers in this thread to describe to me exactly how a mistake like this gets reasonably made. I keep getting replies saying that there are good reasons why something like this would happen, but nobody has actually given me one.

There are no confounding factors in this case that invalidate the freshman physics calculation that tells you that this building design will melt things or light them on fire. There is no excuse for not doing said calculations when designing a curved glass building, especially when you already made the mistake on a previous building you designed. And there is no conceivable way you thought you mitigated the problem while actually failing to do so in a reasonable fashion.

As another person with rather more knowledge of architectural design processes pointed out in great detail on this page, the group that designs the facade is not the same as the group that models its behaviour under certain conditions, which is not the same as the group that makes the final decision on the tinting of glass panels, which is not the group that erects it to surprisingly imprecise tolerances.

We're not talking about something obvious like Linux vs Windows. We're talking about something which is superficially obvious like integers can exceed X digits ...where the actual integers are the product of a decidely non-obvious set of calculations intended to generate low values but affected by code several specialist teams have commit privileges to.

We're talking about a very slight degree of concavity on part of a large facade focuses the sun's rays for two hours a day, two weeks a year provided the surface material is sufficiently reflective and London has less than its median 68% cloud cover for September. And we're also talking about a building that isn't finished yet: here's an architect's rendering of the offending facade that appears to have ridges the current sheer shiny wall doesn't http://www.google.co.uk/imgres?um=1&safe=off&client=firefox-...

It's not like either glass facades with some degree of concavity are unusual or being singed by them is common. Unlike database errors those stupid developers keep coming up with.

n.b I don't even like the building; it's fugly from where I'm sitting. I'm going sunbathing there tomorrow though.

Except it's not "non-obvious" at all, but highly obvious.

The people who designed the facade are not the same group that models the behavior under certain conditions? Well, that's apparently stupid, and somebody in the first group should have said, hey, where does this thing focus sunlight?

At some point, somebody drew a giant curved reflective surface and failed to consider what it would do to the sunlight. That is dumb. I don't see any way around that.