Over the weeks since the G20 riots and the death of Ian Tomlinson, after the damning phone- videos and 200 other allegations against the police, evidence has been heard by the Home Affairs Select Committee. Today we shall learn what it made of it.

There has been plenty said. Denis O’Connor, the chief inspector of constabulary (whose own report is due in autumn), fiercely condemned the “kettling” tactics of imprisoning even innocent passers-by for hours, and excoriated officers who hid their ID numbers. Two policemen have been suspended; numerous claims of abuse have been made by protesters who say they offered no provocation. The MP Tom Brake told the committee that he saw plain-clothes officers slipping through a cordon and that witnesses said to him that the officers incited the crowd to throw bottles. Harsh treatment of journalists caused, as usual, particular outrage in the media.

Countering this, some of the senior police evidence to the select committee sounded smug: as one reporter exasperatedly put it, they said “the policing was good, the protesters were bad, the media were very bad and our few bad apples will pay the price”.

For me there are two contrasting threads to tug at, and they are both concerned with police training. It presents particular challenges because police work is bound to be a mixture of static boredom (waiting, guarding, sifting evidence, recording) and high-adrenalin excitement (chasing, arresting, disarming). In this respect it is like soldiering; but police should not be soldiers. They have no permanent enemy, they are not at war with us. As Mr O’Connor said, the Queen’s Police Medal bears the simple inscription “To guard my people”.

Yet they do have to respond robustly to any of the Queen’s people who suddenly make themselves enemies of others. It is tricky: it calls for a dual identity inside every copper. When you see it working nicely, as I did for The Times when I followed a Saturday night beat amid drunken clubbers, it is fascinating to watch the transitions. One minute PC Rafferty was tolerating being hugged by staggering hen-parties, chatting with a landlord or standing statue-still watching from a distance as a club queue grew impatient. The next, he was running up the road to stop a nasty fight, not knowing what weapons he might meet.

And it was observable how quickly — after such a surge of adrenalin — this experienced officer brought himself back under control. His heart clearly slowed faster than mine did. In other situations I have seen that self-calming take longer: the restless, twitching nerviness of the rookie policeman, eyes darting, hand on baton. It is understandable. When you feel threatened, every muscle stays ready to lash out. Even office workers know that (to the detriment of their digestions) and on tough streets the response is visceral. Remember that, however flawed and sometimes disgraceful the police response was, it is a fact that the G20 protest crowd included strong, angry people willing to break windows, throw missiles and set fires.

The TV footage brought out two things. One was — of course — the awful irrationality of some police lashing out at Tomlinson and others. The second was that aerial view: a phalanx of yellow-jacketed police, a small triangle, hemmed in by a dark surging sea of protesters. I reflected how hard it must be to behave properly in such circumstances, and how much practice it needs.

So what are the two training problems I spoke of? The first concerns the elite Territorial Support Group (TGS). The Times sent me out with them too, on a stop-and-search mission. They were an impressive bunch, and their training is intense: every five weeks they go down to Gravesend for refresher sessions. They learn to face petrol bombs, prison riots, terror attacks. They practise with riot gear, helmets, shields, taser guns. Before our night’s deployment they were severely lectured on the public’s rights and how all actions must be PLAN — “Proportionate, Legal, Appropriate and Necessary”.

However, it must be said that the TSG has attracted its share of angry allegations, 200 a year or so; and there is a strand of professional opinion which makes a case against precisely this sort of “paramilitary policing”. The argument runs that their equipment and attitude make violence implicit wherever they go: the very training creates a self-fulfilling prophecy. This is not proven — the TSG have often policed events which didn’t turn violent. But that may be one problem: over-specialist, over-adrenalised, heavily equipped paramilitary-styled units who may, or may not, make violence more likely but who certainly attract suspicion.

The other training issue is the precise opposite. It is said that one conclusion of today’s report will be that undertrained, inexperienced officers should never have been sent to a protest front line like this. We learn that ordinary officers have lately had their training in public disorder cut from four days a year to two — to save resources. Some of those on the streets on April 1 probably hadn’t had any. Certainly Commander Bob Broadhurst told the select committee that the “vast majority” of his officers had never faced anything as violent as G20 day. It is an awful echo of the Army abuses in Iraq, caused partly (informal army websites surmise) by the abolition of routine prisoner- handling courses.

So perhaps some of the police who lashed out were just terrified young men pitchforked without advice into an urban battlefield. And perhaps some of the others were edgy wannabe soldiers, for whom battle was nastily exhilarating. And we don’t want either: not PC Panic, not the Terminator. Policing civilians is a subtle science and a woefully unresearched one.

Maybe we should listen to the surgeon Professor Jonathan Shepherd of Cardiff. Provoked by the injuries he has to treat in maxillofacial clinics, he has long banged the drum for rigorous evidential research into what kinds of policing and training actually work best. He advocates the foundation of University Police Schools, bringing together disciplines from psychology to statistics, law, geography and economics.

A scientific approach would make a change from media extrapolation, self-defensive obfuscation by senior police, and political grandstanding (handcuff-waving, cashpoint- marching). It could damp down the tendency to lurching, veering panic Home Office decisions.

If the G20 day leads to thoughtful- ness on the subject it may do us, and PC Everyman too, a better service than the summit itself ever did.

Times Recommends

• Bank charges hit us for being human
• Sorry, Father, thou still shalt not steal
• The shed where Santa’s failures go to die