The current fashion for splitting the decks of wide bridges mirrors the architectural preoccupation with breaking down the bulk of large building complexes. What might be one large building can become a township of smaller structures, bringing the landscape into the design. With bridges, the need for balance tends to provide less scope for complexity, though for example Zaha Hadid's leaping architectural forms lends themselves particularly well to bridge projects, and she is responsible for several. Perhaps more typical might be Nicholas Grimshaw's bridge at IJburg outside Amsterdam, which must carry road, rail and pedestrian traffic side by side to a new island settlement. The architects saw the double enclosing arches of their bridge as rooms, designed the facetted joints of the arches to resemble those of marine crustacea, and split the deck into several thinner strips so as to provide a visual link with the water beneath. They also decided to link the two arches with a bowsprung under-arch at the central linkage point, so creating the effect of a sinusoidal wave. These were very architectural decisions, taken early in the competition process. The competition itself was unusual, because previously the Netherlands had pursued a policy of reclaiming land to form polders contiguous with the mainland: consequently bridges were of little import. At IJburg, a different policy was adopted: the new settlements - effectively comprising a satellite town for Amsterdam - would be created on a number of linked islands. Water automatically creates the edge condition, and the bridge (and subsequent bridges to a broadly similar design) provide the necessary markers at the points of entry.

But what form should such markers adopt? By chance I was present when two English architects - both of whom had won competitions to design pedestrian bridges - met each other at a party. Architect A congratulated the younger architect B on a recent victory. "Very interesting," he said, "there are only a certain number of ways to design a bridge. And yours was one of them."

Quite a put-down. Yet it reflects the undoubted fact that - despite extraordinary technological advances in recent years - engineers and architects perhaps have less freedom in bridge design than might at first appear. You cannot defy the laws of physics, though you can always - as Santiago Calatrava is adept at demonstrating - delay the point at which the load finally meets the ground by transferring it elegantly through sub-structures. Few can do this with the elan of Calatrava. More usually, bridges treated in this way tend to end up over-complex. If there is a rule, it is to pare down the design. Which is why so many bridges look like other bridges. It is rare to find a genuinely original solution, but it happens, and a rich crop has emerged in Britain in recent years. For instance, Wilkinson Eyre's S-curved, tilting-masted cable-stay pedestrian footbridge in London's docklands - one half of which, just to make the whole thing even more ingenious, rotated open for shipping. Or Hodder Associates' hyperbolic paraboloid pedestrian bridge in Manchester, like a power station's cooling tower placed horizontally, the shape of which allowed the footway inside the glazed tube to slope between the different floor levels in the buildings on either side. Or indeed the Royal Victoria Dock footbridge, out east in London's docklands, by architects Lifschutz Davidson and engineers Techniker (led by my co-author Matthew Wells). There the concept of a transporter bridge is revived and combined with a dash of structural audacity: a "Fink Truss" system is inverted, and in the process subverted, to provide a highly articulated profile very necessary in the flatlands of East London.