Government Communications Headquarters (GCHQ) is the largest intelligence agency in Britain, a rank it has held since its predecessor expanded to confront coming war in 1939. Its principal task is "sigint" (signals intelligence): providing clients with information derived from intercepting, or spying on, the communications of others. Every detail of GCHQ's size, mission and methods was unknown to the public, press and most of parliament for more than half of its 70 years so far. The agency's capabilities - and, importantly, the limitations of these capabilities - are little understood.

More difficult still to foresee is the role that the new surveillance behemoth inside GCHQ's doughnut-shaped headquarters in Cheltenham may come to play as our lives and social being migrate fully into electronic form. In Britain as in the United States, the physical means to tap, track, store and assess this data are already embedded in secretly spliced optical fibre cable connections that loop around national telecommunications systems, pulling every sort of communication into an unseen nebula of data warehouses. But is this activity merely a large and super-secret version of Facebook or Google, "a vast mirror [that] reflects the spirit of the age", as this book concludes?

Richard Aldrich, an accomplished cold war intelligence historian, has taken a decade to produce the first substantial account of what is known about the agency, and what can be gleaned from recently released official archive material. He charts how, by 1964, GCHQ's demands and hidden financial allocations exceeded the entire cost of the Foreign Office. Its managers lobbied for a string of ambitious and costly projects: a nuclear-powered, aircraft-carrier-sized spy ship (never built); a small force of sky-sweeping Nimrod spy planes (flying from Lincolnshire now); and a spy satellite, Zircon (which never left the ground).

After decades of darkness, "GCHQ was unmasked in the summer of 1976", the book acknowledges, by a "path-breaking" article entitled "The eavesdroppers", which I wrote for Time Out. The agency's unwilling transition to public awareness was consolidated by the subsequent "ABC" Official Secrets Act trial of 1978, directed at myself and two others. Aldrich continues: "together Duncan Campbell and James Bamford [who wrote the first book about GCHQ's American equivalent, the NSA] have confirmed a fundamental truth: that there are no secrets, only lazy researchers". That's very kind, but I can't agree. I'll come back to that.

Behind and in front of GCHQ, there are the collectors and the recipients. The collectors use dangerous, costly means of bringing in signal intercepts: reconnaissance aircraft, special submarines, covert and uncomfortable listening stations in embassies and diplomatic missions. The recipients are the select few with clearance to see the "sensitive compartmentalised information" that is generated.

During the 1960s and 1970s, military risks were taken and important foreign policy decisions managed to support the demands of the intelligence collectors. Among these events was the manipulation of decisions about the British military presence in Cyprus solely to obtain convenient real estate for GCHQ and radar bases, and the deportation of the Chagos Islanders en masse from their homes on Diego Garcia. As Aldrich puts it, "the sigint tail had begun to wag the policy dog".

Cold war tales of derring-do are recounted here. But, however exciting, these serve to mask how much of the capability of sigint has been turned inward, rather than out to face the USSR or other conventional enemies. The "D-notice affair" of 1967 was about the wholesale collection of all overseas telegraph messages and their delivery to GCHQ. Before controversy subsided, GCHQ was busy erecting, near Bude in Cornwall, receiver dishes that would track and copy all communications passing through western communication satellites. A later project involved the construction of an £18m tower in Cheshire that tapped British Telecom's network to intercept communications with the Republic of Ireland.

The Bude scheme was the start of the English-speaking allies' Project Echelon, comparable, Aldrich suggests, to today's Google Alert system, which constantly scans the internet for new additions. This is an ingenious comparison, but it omits a critical point of divergence. Google, even though it often overreaches itself, collects what is placed in the public domain. The sigint collectors are scanning and storing the entire private domain of communication, under questionable authority at best, and certainly without accountability as it is normally understood.

Over east London now, as you are reading this, a sigint collection plane is likely circling at 10,000 feet above Canary Wharf, scooping up the capital's cellular networks, reportedly attempting to voice-match mobile telephone calls made in the area to a bomber back in Britain following training with the Taliban. If such activity nets those who plan harm on the City streets effectively, all may appear well and good. But how are the hundreds of thousands of others whose communications are collected to be protected against impropriety, or error, or worse?

Aldrich notes that GCHQ "wrestles with some of the most troubling issues of our time". However, the agency is not so much a mirror as it is an advocate for a world of less privacy and more convenience to the collectors. During the 1990s, along with the NSA, it led a campaign to prevent cryptology being used for email. Both agencies have assiduously sought to bypass telecommunications regulations and standards, undermining privacy and security while preserving access.

One part of this project has been the UK's current law, the Regulation of Investigatory Powers Act (Ripa) of 2000. It would have been helpful here to have had considered the legislative history and present effect of Ripa, and how its arcane language may have been drafted to smuggle past parliament GCHQ's plans and intentions for the future of privacy on the internet.

The lack of properly informed debate about how far the interception of information on the internet has gone, and what it may mean for the future, is not just the fault of lazy researchers. Accessible information on this subject generally does not exist, and where it is unearthed it may be suppressed, as happened when whistleblowers exposed a network of secret rooms inside US internet exchanges where the NSA had been wired in to the fibre-optic cables, unlawfully. (As a US senator, Barack Obama voted to grant the telecommunications companies immunity from criminal prosecution for violations of US citizens' constitutional privacy when the firms handed over communications records wholesale to his immediate predecessor.) In Britain, no similar whistleblowers have emerged.

A great deal of essential and important inquiry remains to be done. Start here.

GCHQ: the Uncensored Story of Britain's Most Secret Intelligence Agency
Richard J Aldrich
HarperPress, 448pp, £30
Duncan Campbell was an associate editor and investigative reporter for the New Statesman from 1978-91