In the wake of three more escapes from control order regimes, Home Secretary John Reid has been coming under fire for failure to keep closer tabs on supposedly dangerous terror suspects. And returning fire with threats to abandon more freedoms in order to keep us free. But a reality check of control orders is long overdue here - these dangerous terror suspects are, the government assures us, not particularly dangerous terror suspects, hence the relatively relaxed supervision regime, and hence there's no need to go into a frenzy about dangerous maniacs being on the loose.

That is, they're sufficiently dangerous for it to be necessary to keep tabs on them, not sufficiently dangerous for them to merit full-scale surveillance, and not sufficiently dangerous to warrant the use of a control and supervision system that they can't walk out of whenever they like. Effectively, although the kind of control order regime implemented for particular individuals varies, each and every one of them allows the subject to walk away at will. At the entry level you have reporting to the local police station at regular intervals, then phoning in from a designated location, possibly checked by slightly flaky voice recognition software, to 'prove' you're there. Then you've got tags that require you to stay within range of the monitoring device, and beyond that you have GPS tags that allow your every move to be monitored. The latter exist, but it's not clear if the Home Office has ever deployed one as part of a control order regime.

All sorts of complicated terms and conditions can be, and are, layered on top of this technological prison without bars. The subject can be forbidden to use the Internet and/or computers or mobile phones, be confined to a specific geographical area, can have their employment or the people they meet subject to Home Office approval, and so on. Virtually none of this, including the barless prison itself, is immediately enforceable, the point if there is one being that any breach of the restrictions can lead to a prison sentence. This can be handy, given that the control orders are allegedly imposed because although the subjects are dangerous/not-very-dangerous there is no evidence that can be presented in order to send them to prison.

So there are similarities to ASBOs, and similarities to the scenarios faced by 'normal' criminals subject to community sentencing or restricted early release. But control order regimes include insane mind games that are not generally present in those other aspects of new model criminal justice.

The mind games and general wackiness of the system are particularly obvious in the case of these three absconders. Two of the three are brothers of Anthony Garcia, who was sentenced to life for his part in the fertiliser bomb plot last month. During that trial one of the brothers, Lamine Adam, was named by 'supergrass' Mohammed Junaid Babar (who gave evidence as part of a bargaining deal with US authorities) as being associated with the bomb plotters. He wasn't charged in connection with this, and the reason given for subjecting the three to control orders was not that they presented a terrorist threat in the UK, but that they were suspected of planning to travel abroad "for terrorism-related purposes."

Control order regimes meanwhile were brought in because the government's previous 'solution' of indefinite prison detention without trial was deemed to breach the European Convention on Human Rights, and because the security services did not have sufficient resources to monitor 'dangerous' suspects loose in the community. Pause briefly to consider the eccentricity of that last claim. Fairly recently the number of terrorism suspects being monitored by MI5 was claimed to be in the region of 2,000, while the number of people subject to control orders would have totalled 17 if they could manage to keep hold of all of them at once. Six are currently AWOL. It is not clear why M15 has the resources to monitor 2,000 people, but lacks the resources to monitor 2,017 people. Nor, if people under control orders are a bit less dangerous than people directly monitored, is it clear why there are 2,000 of the more dangerous ones but only 17 of the less. What strange statistical model does this distribution conform to?

Whatever, we have three people who, if they're dangerous at all, are dangerous because they want to leave the UK and be dangerous somewhere else, so we control them with a system that they can walk out of any time they want, and the only thing impeding their progress out of the UK (if that's what they want to do) is the fact that we've confiscated their passports.

Now, let's suppose for a moment that actually they didn't want to leave the UK, go to Basra and blow themselves up or similar. Or not necessarily that strongly, anyway. The Adam brothers had their control orders imposed in February 2006, the third man in July 2006), and a control order is a sentence without limit, so they've already spent around a year in a form of restrictive internal exile, with no idea when it will ever end. Disappearing would surely begin to look attractive, would it not? The deliberate mind game aspect of control orders and tagging and monitoring regimes is that the subject thinks that breach will lead to punishment, and will therefore conform. But the accidental mind game of control orders is that the regime is the punishment. Prison might well be more attractive, but once you'd served that they'd probably put you right back under a control order, so the only thing stopping you running is the dream that one day it might all end. Those control order subjects with families do have some incentive to conform, but not those with little or nothing left to lose, surely.

On the other hand, consider the situation if the security services were right about these men, or about some of the other control order subjects. What we would have in this case is determined individuals with a desire to plan and execute acts of terrorism. They would do so, one might surmise, by associating with other potential or actual terrorists, acquiring information, contacts and planning operations. That's what MI5 thinks the 2,000 more dangerous suspects are up to, anyway, which is why they're watching them. One might observe that watching people ought every now and again to lead to dangerous controllers abroad, if the real world works like the MI5 script for it. Mysteriously, however, connections of this sort identified so far have tended to be identified by the Pakistani or Moroccan secret services, or by the odd supergrass held by the Americans.

So one does begin to wonder about the efficacy of MI5's observations. But presuming your initial suspects are genuine, the methodology is sensible enough. You watch them for contacts, bug them, watch plots unfold, then at the optimum time (N.B. this is not when some crazy person tells you about some impossible bomb that will destroy East London) you spread the net and pick up everybody, shadowy al-Qaeda controllers included.

Back within range of the beeping boxes (actually, it seems this week's three weren't even tagged), you can see how this approach works. You've got your dangerous suspects who know, er, that you're not watching them because they've been given a control order because you say you don't have the resources to watch them full time. Obviously, as you suspect them, they'll suspect that secretly you might be watching them a bit anyway, or every now and again, so if they're serious about terrorism they'll be careful. But, consider the insanity of the rest of the deal, they might just reckon you're full of shit, not watching at all, so they're perfectly safe to plot the big one and then scarper just before it goes off. Mind games again, but if you are full of shit, you should worry about them figuring this out and stopping believing that you're watching.

So far, no control order subject has committed a terrorist atrocity of any size or description after absconding. Several control order subjects have ended up needing psychiatric attention, a couple of Algerians have concluded that going back to face the tender mercies of the Algerian secret service is preferable to life at the bleeding edge of Home Office criminology, and others have just opted to vanish. Defending himself this week, John Reid said that he had consistently said that control orders were not the best option, or even the second best option, for dealing with terrorism suspects. But he undersells himself here - clearly, they are not even a rational option, and whoever's responsible for them (Charles Clarke, based on an idea by David Blunkett) wants locking up. ®