Many authors including Joel wrote a lot about the value of concentration, or being in 'the zone' as in "a mode of cramming out code" - which it is rather hard to enter and easy to exit. Quoting Joel - during his work, a programmer has to remember a whole bunch of class, variable and function names, and therefore, context-switching is considered harmful. Context-switching is inevitable in open space, so advocates of this idea recommend private rooms with doors that close.
There are, however, situaitons in which this can be good. Scientists and algorithmists I talked to, often report that they come up with greatest ideas not at the time of concentrations and that random input from peers and thinking about many problems, and switching problems in particular helps them produce better results.
During the time of integration of a large system of many components, human-to-human interaction is key to success, and removing any boundaries is crucial. Sometimes I used to move my whole team to a large room for a "war integration" effort.
Another time when open space can be useful is the initial stage of a new project, when people try to collect as many ideas as possible, before the planning stage. It works like continuous brainstorming, where anybody can pick up and develop thoughts of his teammates.
Some managers might seem open space as a management tool, where they can instantly see what everybody is doing. I believe this problem should rather be addressed by good managers, who maintain the workplan, priorities and time estimates continuously.
1. It is better to avoid open-space for software teams. It might be tempted, but unlikely helpful in long-term.
2. Provide your employees with good managers, who will let them focus on their tasks.