X-Clacks-Overhead
Amusingly, that's not true. The only cookie they send is Google Analytics, which has zero value to the user. The site works fine with it blocked.
Do you want to know how many human years my last company had to devote to regulation? We could have built a hundred startups with all that effort.
I'm not saying GDPR right to be forgotten and data dump/portability isn't important, but it comes with a steep cost that everyone pays everywhere. So much time and money was spent on it. Easily billions of dollars.
And the cookie stuff? How useful has that been?
ePrivacy and GDPR compliance are cheap. Trying to rules-lawyer them to keep illegal business models going, while dodging regulatory scrutiny, is expensive.
ePrivacy and GDPR compliance are cheap.
Try running a business that has to maintain GDPR compliance and KYC / AML / FINRA compliance. That is not cheap.
Building compliance is not building for your customers direct asks and requirements. Especially software that does not originate in the EU. How many startups are building data export to comply with data export regulations?
I spent nearly a year plumbing through complex microservices to satisfy GDPR at my last company. We collected an enormous amount of PII and KYC data from payments processing, and there were so many downstream services impacted. And I was just one engineer from amongst dozens of impacted teams that had to deal with it.
Regulatory compliance is not free.
Regulatory compliance is frictionful.
I'm not saying regulation is bad, but that it is a cost of doing business and a tax on engineering. Especially for startups looking to go toe to toe with bigger incumbents that have already paid for compliance and that can afford to pay fees to ignore compliance to go fast.
GDPR-compliance in a greenfield project is cheaper than dirt, up until someone makes a GDPR request, at which point it's slightly more expensive than dirt because you had to take 15 minutes out of your day to satisfy the request. By your third or fourth GDPR request, it's perhaps worth taking time to implement an automated flow, but having that many customers is a lovely problem to have!
But my point is orthogonal to liking the regulation.
GA was the only way to get a simple page count view without setting up a database or a backend system before we switched to serverless cloud step function lambda craziness.
Seems like a passion project, and they just wanted to know if their work was used. I give people the benefit of the doubt on all of this when it's a small site.
I figured this was one of the best ways to do it. That way I'm letting people that were significant to me live on forever, one random HTTP response header at a time.
$ curl https://xeiaso.net --head | grep clacks
x-clacks-overhead: GNU Satoru Iwata
This is most noticeable in his caricatures that became characters that became badasses over multiple novels; the Watch has a few of these, but there are others.
And to what end? To make less money than their moms do in internal medicine?
It really seemed like Pratchett knew something of this niche cultures, way more than I expected.
It really seemed like Pratchett knew something of this niche cultures, way more than I expected.
He was definitely an early adopter of the internet, (and e.g. very active on alt.fan.pratchett), so that's no big surprise.
In Terry Pratchett's science-fantasy Discworld series, "The Clacks" is a network infrastructure of Semaphore Towers, that operate in a similar fashion to telegraph - named "Clacks" because of the clicking sound the system makes as signals send.
Surely named "Clacks" because of the clacking sound the system makes.
https://en.m.wikipedia.org/wiki/Abraham_Niclas_Edelcrantz
Also UK used a system close to that. And a lot of countries along Europe developed their networks with different signaling devices.
https://en.wikipedia.org/wiki/Chappe_telegraph
The stations were more elaborate and there is even a recorded instance of a secret signal being passed on illicitly:
https://blog.franceinfo.fr/deja-vu/2017/10/10/le-piratage-du...
The early history of data networks https://archive.org/details/earlyhistoryofda0000holz
https://terrypratchett.com/explore-discworld/the-modernisati...
[...] header that can be transmitted from server to server [...]
How so? In HTTP, there's always one client and one server. Am I missing some way to make this sticky or self-propagating, e.g. browsers or other clients that will cache received headers and then send them to other servers?
There are also browser extensions, which show when a website broadcasts the "X-Clacks-Overhead" - header.
One day I noticed that it disappeared, but then it returned, so someone on the inside cared and brought it back, that made me smile :)
https://www.shodan.io/search/report?query=x-clacks-overhead+...
For some reason, a lot of honeypots are also using that header so I filtered those out. The number of services has slowly increased over time:
https://trends.shodan.io/search?query=x-clacks-overhead+-tag...
Further down the info, we see 28,587 (almost the same number as above) HTTP titles are "Gargoyle Router Management Utility" - which is an opensource variant of the OpenWRT world which patches the code to include the Clacks header.[2]
I'm going to conclude that there's a direct correlation in this data (it all being one and the same endpoint/device pattern) and that 30,000 KT Wifi hotspots across South Korea have their management UI open on the public interface and not locked to the internal network or a VPN, etc. running this Gargoyle patch.
[1] https://en.wikipedia.org/wiki/KT_Corporation
[2] https://github.com/ericpaulbishop/gargoyle/blob/master/patch...
< HTTP/2 301
< server: nginx
< date: Sat, 05 Jul 2025 13:36:11 GMT
< content-type: text/html
< content-length: 162
< location: https://www.mozilla.org/
< strict-transport-security: max-age=60; includeSubDomains
< x-backend-server: TS
< cache-control: max-age=3600
< via: 1.1 google
< alt-svc: h3=":443"; ma=2592000,h3-29=":443"; ma=2592000
Edit: Nope. I was wrong, if you follow that 301 it does:< x-clacks-overhead: GNU Terry Pratchett
Perhaps something like IPv6's Hop-by-Hop Options can be used to pass names with every packet?
Or, even better, we can use LoRa repeaters for something close to the actual clacks network.
I need more time and motivation to make a full network though.
(I used to administer a laser link. go on, ask me why they aren’t very popular)
I spent a lot of time working out how to create low powered laser transducer, capable of working on something battery powered.
This is my favourite part; very real.I think you're right; I suspect Terry would have been tickled by the header, but if there were any physical world implementations I think he would have been overjoyed. One of my favourite Terry stories is of him making his sword, which feels similar.
i back-of-napkin'd a whole packet-over-laser relay system based conceptually on the clacks that'd give every room/station its own serial-interfacible (up|down) link. you could link buildings out of windows and stuff. horribly impractical and prohibitively expensive, but the kind of thing that could only happen in a university on-campus environment.