When Blocking Went Blunt

July 2006. Days after the Mumbai train bombings killed more than 200 people, the Department of Telecommunications sent the country's ISPs a letter: block this list of specific sites — seventeen of them, a handful allegedly carrying extremist content. A targeted order. Surgical intent. The kind of thing a state assumes it can do to a network, because it has always been able to do it to a printing press or a radio tower.

Then the order hit the actual infrastructure, and the actual infrastructure could not do surgery. ISPs at the time blocked at the domain level, not the URL level. So to kill a few pages hosted on Blogspot, they blocked all of Blogspot. And Typepad. And GeoCities. Every blog on those platforms — hobby recipes, travel diaries, somebody's cat — went dark across India, all to bury a list most people never saw.

The bloggers noticed before the government explained anything. It started around July 13; by the 15th people were phoning their ISPs and being told, on the record, that yes, the sites were blocked, there's a letter from the Ministry. One caller reached Spectranet's call center and got a flat confirmation. You could still publish to your blog through Blogger.com — you just couldn't see it at Blogspot.com. The write path worked; the read path was gone. A censorship regime so poorly aimed it forgot to close the back door.

And here's the part that should have been the headline, then and now: the fix was trivial. Within days, bloggers were routing around the block with proxies — pkblogs, web relays, the usual samizdat of the early web. Others filed Right to Information Act requests to force the government to admit what it had done. The crudest form of the block lasted about a week. The technology to censor was blunt; the technology to evade it was a browser tab.

Twenty years later we've got a grander word for state control of the network — the splinternet — and it usually means something more deliberate than 2006: data localization, national stacks, parallel DNS, fragmentation by design. The India block wasn't that. It was cruder and more honest — a state discovering that "the internet" is not one controllable thing but a stack of layers it only half-controls, and that control applied at the wrong layer produces collateral damage out of all proportion to intent. 2006 didn't invent the splinternet. It prototyped the failure mode underneath it: block a URL, lose a platform; block a platform, radicalize a thousand apolitical bloggers who just wanted their site back.

The comfortable read is that this was incompetence — old routers, unsophisticated ISPs, a government that didn't understand the web. That read is wrong, and it's the dangerous one. In 2006 the bluntness really was baked into the pipes: packets don't carry the context a censor cares about, so the state wanted to block a sentence and the infrastructure could only block an address. But that gap did close. DPI, SNI inspection, national firewalls — the modern stack can block a single URL, a single Wikipedia article, a single video. Precision arrived. What didn't change is that states still reach for the sledgehammer when it's cheaper, or when the collateral damage is the point — over-block a platform and you don't just bury seventeen pages, you teach a million people to stop looking. The bluntness stopped being a limitation and became a choice. That's the worse version, and it's the true one.

What the episode actually proved is uncomfortable for everyone. The internet is not censorship-resistant by design — that's the founding myth, and it's a myth. It was censorship-resistant by neglect: the tools were crude, the operators were sloppy, and the exits were cheap. Every one of those conditions has since been engineered away. The commons stays open exactly as long as the infrastructure underneath it stays open, and the infrastructure belongs to whoever holds the routers.

And here's the part that should keep you up: the exits consolidated too. In 2006 evasion was a proxy and a browser tab — samizdat anybody could run. Today's VPNs and proxies ride the same handful of cloud providers a state already leans on, and the app stores that gate those apps answer to the same pressure. Censorship migrated up into a few controllable chokepoints — and so did escape. That symmetry is the whole story: cheap exits stopped being cheap because both sides climbed the same ladder into the same small number of hands.

The bloggers won that round with a proxy and a middle finger. I wouldn't bet on the rematch being that easy.