On City Comforts they make the case that Daniel Libeskind's abominable extension to the Victoria and Albert museum could be made to fit with the museum and South Kensington in general. To support this they link to this Good City post on traditional neighborhoods and modern architecture, which argues that the only thing necessary for a modern building to fit in a traditional neighborhood is that the site plan be well integrated into the public space. As evidence for this they show a picture of a modern house in Lincoln Park, Illinois. The house in that picture actually has qualities that the Libeskind addition doesn't have, mainly that it has several scales of geometry symmetrical with its context. The walls of the modern building are symmetrical with the walls of the historic buildings, and all buildings are thus linked together into one fabric. The windows are different, but that is the outcome of adaptation to changing needs, which requires new scales of geometry.
Integrated site plans are a necessary but insufficient condition of good urbanism. The idea that you can make a whole out of an anything goes architectural approach was rejected even by nihilist-leaning Rem Koolhaas. In complex systems symmetry is found because it is the most physically efficient process. Remembering that the definition of symmetry is a preserved structure after undergoing a transformation, we can more clearly define what kind of modern architecture is good for a historic neighborhood. A new building fits into a historic neighborhood if its geometry can be derived from an old building with the fewest possible transformations. That is to say, all reusable scales are reused, and the modern building has all the useful new geometry that we require in our time. What makes a neighborhood historic are those reusable scales and not the age of the buildings. The building in Lincoln Park reuses the wall scales of its neighborhood. Now try picturing it with bright rainbow walls and neon. Does it fit anymore? No more than does the Libeskind extension.
Here is a similar building in Paris' 19th borough. Its site plan is as urban as could be made, but it is definitely not Parisian. It has no scales in common with its neighborhood. This happened despite the fact that the majority of real estate in Paris is modern. Modern Paris goes unnoticed because it is symmetric enough with its neighborhood that it contributes, or at least doesn't take away from, the city as a whole.
Comments
If one is concerned about symmetry then maybe the solution would be to hire Gehry to do a building on the other side of the street. No?
If you were to do that, you would then have two symmetrical buildings that clash with the greater context, instead of only one. You would also have to remove part of the context for that additional building. Eventually you would have to remove the whole neighborhood to match the Libeskind extension. How does that make sense?
That would be a good idea if one was building a new city from nothing, or building in a city with a lot of holes, let's say Houston. You could tell Gehry and Libeskind to go wild, and the result would be a highly symmetrical environment that would happen to be the grooviest place ever built. Twenty years later it would probably be a UNESCO heritage site. And what if someone wanted to build an exact replica of the Victoria and Albert museum in that city? That would break the symmetry and crash the street line.