The Internet Galaxy and the Network That Became Society
Manuel Castells's The Internet Galaxy is an early map of the internet after it escaped the laboratory and before platforms, phones, clouds, and AI agents made networked life feel ordinary. Its value now is not prophecy. It is diagnosis: the internet did not sit on top of society. It became one of the main forms through which work, identity, markets, politics, media, and surveillance learned to organize themselves.
The Book
The Internet Galaxy: Reflections on the Internet, Business, and Society was published by Oxford University Press in 2001. Oxford Academic's record for the paperback ISBN lists the book at 292 pages and gives the online ISBN as 9780191698279. The University of Cambridge sociology research listing describes the book as a study of internet evolution and social influence, with particular attention to the internet's development from 1962 to 1995.
The title answers Marshall McLuhan's The Gutenberg Galaxy: Castells is asking what kind of civilization emerges when the dominant communication environment is no longer print alone but networked digital connection. The chapters move through internet history, internet culture, e-business, work, politics, privacy, multimedia, geography, and the digital divide. The result is less a narrow technology book than a social anatomy of a new operating environment.
Castells's basic claim is not that the internet magically changes everything. It is that networks amplify and reorganize forces already present: capitalism, state power, academic research, hacker culture, entrepreneurship, media, social movements, and everyday communication. That is why the book still matters. It treats the internet as infrastructure, culture, institution, and political economy at once.
Internet Culture as Institution
The strongest early move in the book is to separate the internet from the myth of a single origin. Castells traces a layered culture: military-funded research, university science, hacker openness, virtual communities, and entrepreneurial innovation. The internet is not born from one ideology. It is assembled from conflicting institutions that learned to share protocols, values, and incentives.
That layered origin helps explain why networked systems can feel liberating and controlling at the same time. Openness, peer production, experimentation, and distributed communication are real. So are commercialization, surveillance, infrastructure dependence, moderation power, and the enclosure of shared activity inside private platforms. The same network can host mutual aid, conspiracy, labor extraction, state monitoring, and market speculation.
This matters for AI because current model culture has a similar mixed inheritance. Research norms, open-source ideals, venture finance, defense funding, cloud monopolies, content scraping, creator labor, and product growth metrics all meet inside the interface. A system can inherit scientific language while being governed by commercial deployment. It can sound like public knowledge while routing attention, data, and dependency toward private infrastructure.
The Network Enterprise
Castells is especially useful on the network enterprise: firms organized less as sealed hierarchies than as flexible webs of suppliers, partners, contractors, platforms, and information flows. The internet did not merely make companies faster. It changed what could be coordinated at a distance, what could be outsourced, what could be measured, and how labor could be modularized.
Read from the AI era, this looks like a prehistory of platform capitalism and agentic work. Once business processes become networked, they can be logged, routed, scored, predicted, and partially automated. The same connective tissue that lets teams coordinate across the world also lets management observe work at higher resolution and lets vendors insert tools into the middle of ordinary judgment.
The network enterprise also changes institutional memory. Work leaves traces: tickets, chats, commits, transactions, dashboards, calendars, documents, customer records, and model prompts. AI systems can now summarize and act on those traces. The firm becomes a machine-readable environment, and machine readability becomes a form of power.
Politics in the Network
The Internet Galaxy was written before the mature social-media age, but it already sees that politics changes when communication networks lower the cost of coordination and publication. Networked politics can bypass gatekeepers, expose official narratives, and create new forms of participation. It can also fragment attention, weaken shared institutions, and make legitimacy harder to stabilize.
The point is not that networks are naturally democratic or naturally authoritarian. Castells is more useful than that. Networks alter the terrain on which democratic and authoritarian forces operate. A movement can scale through networked communication; so can harassment, fraud, propaganda, and synthetic consensus. A state can be challenged by distributed publics; it can also use networked data to classify, predict, and preempt them.
That ambivalence is now basic to AI politics. Generative systems sit inside networked media environments that already reward speed, personalization, outrage, imitation, and metrics. They do not enter a neutral public sphere. They enter feedback systems where visibility becomes evidence, repetition becomes familiarity, and coordination can be automated.
Privacy and Legibility
The privacy chapters have aged in a clarifying way. Castells wrote before smartphones, real-time bidding, data brokers at today's scale, cloud AI, and consumer chatbots with memory. Yet the underlying pattern is visible: networked life produces traces, and those traces become attractive to firms and states because they make people administratively and commercially legible.
Privacy is not only secrecy. It is the ability to move through life without every gesture becoming an institutional input. When communication, shopping, work, search, mobility, entertainment, and friendship pass through networked systems, the person is reconstructed as a profile, a risk, a market segment, a user, a target, or a behavioral stream.
AI sharpens this because traces are no longer only stored or searched. They can be interpreted, summarized, inferred from, and acted upon. A dossier can become a recommendation, a denial, an automated suspicion, a synthetic memory, or a conversational assistant that remembers a version of the user. The old internet made social life more legible. The AI internet makes that legibility operational.
The AI-Age Reading
The book's most useful AI-era lesson is that artificial intelligence is not arriving from outside the network society. It is being built from it. Training data comes from networked culture. Cloud compute sits in networked infrastructure. Deployment depends on platforms, APIs, browsers, phones, enterprise software, app stores, and search. Governance depends on institutions that already struggled to govern the internet.
This is why the agentic web is not just a new user interface. If software agents browse, buy, negotiate, schedule, summarize, and decide through networked services, then the internet becomes less a space people visit and more a substrate that machines operate on behalf of people, firms, and states. The human user becomes one actor among automated actors, platform rules, ranking systems, trust layers, and vendor-controlled memories.
Castells helps keep the scale of that shift visible. The internet was never merely a medium for messages. It became the coordinating infrastructure for business, politics, culture, and identity. AI systems now learn from that infrastructure and feed back into it. The danger is not only smarter tools. It is recursive social infrastructure: networks produce data for models, models reshape network behavior, and the changed behavior becomes new evidence about reality.
Where the Book Needs Care
The book's age is visible. It predates platform monopolies as we now know them, smartphones, algorithmic feeds, cloud concentration, large-scale social-media moderation, the surveillance advertising stack, deepfakes, and generative AI. Some of its examples belong to a web that feels almost gentle compared with today's integrated platform environment.
It can also lean toward grand synthesis. Castells is at his best when mapping institutions and network forms; readers looking for the lived texture of platform labor, racialized surveillance, content moderation, feminist data politics, or ecological extraction will need companion texts. The internet galaxy is not only a structure of flows. It is also built from mines, cables, warehouses, moderators, data centers, outsourced workers, payment systems, legal regimes, and unequal exposure to harm.
Still, those limits do not erase the book's usefulness. They identify where the map needs new layers. Castells gives the skeleton of the network society. Later books supply the nerves, injuries, incentives, and hidden labor.
The Site Reading
The Internet Galaxy belongs in this catalog because it explains the world AI inherits. The model is not floating above society as pure cognition. It is trained on networked records, deployed through networked platforms, priced through networked markets, governed through networked institutions, and trusted through interfaces that make enormous systems feel personal.
The practical lesson is to stop treating networks as background. Every chatbot, agent, recommender, search answer, automated workflow, identity layer, and civic dashboard sits inside a social architecture that decides what can be seen, who can act, who is measured, who is exposed, and who can contest the result.
Castells's internet was already a galaxy: not a tool but an environment. The AI era does not replace that galaxy. It animates it. The question is whether people and institutions can still see the network clearly enough to govern the machine intelligence now moving through it.
Sources
- Oxford Academic, The Internet Galaxy: Reflections on the Internet, Business, and Society, publisher listing, ISBNs, chapter list, and publication metadata, reviewed May 19, 2026.
- University of Cambridge, Department of Sociology Research, The Internet Galaxy. Reflections on the Internet, Business, and Society, publication details, summary, and translation note, reviewed May 19, 2026.
- Google Books, The Internet Galaxy, bibliographic listing and publisher description for the 2001 Oxford University Press edition, reviewed May 19, 2026.
- Gerard Goggin, review of The Internet Galaxy: Reflections on the Internet, Business, and Society, Media International Australia, vol. 108, no. 1, August 2003, pp. 174-176.
- Shyamkrishna Balganesh, "Freedom and the Internet: A Review of The Internet Galaxy by Manuel Castells", International Journal of Communications Law and Policy, issue 8, winter 2003/2004.
Book links are paid affiliate links. As an Amazon Associate I earn from qualifying purchases.