Christian Huitema

Working on that Internet thing...

https://www.privateoctopus.com/about.html

Web: https://www.privateoctopus.com/about.html
Github: https://github.com/huitema

Public Key

npub1sysxmkze3h9k7wculavyuz8ms5wa2e64sm52j0802al6k0dwm0pshv4k4f

NIP-05 Address

[email protected]

Profile Code

nprofile1qqsgzgrdmpvcmjm08vw07kzwprac28w4va2cd69f8nh4wlat8khdhscprpmhxue69uhhyetvv9ujumt0d4hhxarj9ecxjmntd6ajz0

Publishing to

relay.momostr.pink

Author Public Key

npub1sysxmkze3h9k7wculavyuz8ms5wa2e64sm52j0802al6k0dwm0pshv4k4f

Show more details

Last Notes

2026-08-22 18:51:02 GMT

- reply

Blocking the GitHub user Claude is not foolproof. It is completely possible to use Claude and not let it claim authorship of the PRs.

2026-08-13 17:43:57 GMT

The proposal to register a Jewish identity in government records in California reminds me of the French police compiling a big file of Jewish residents in occupied France. The Gestapo found that very helpful. Of course, government files compiling racial origin, immigration status or sexual orientation would be just as helpful for the next incarnation of the Gestapo.

2026-07-08 05:21:53 GMT

- reply

I was expecting popcorn. I was not expecting Jacob Applebaum to post a precise description of how a change made by NIST in the final stages of ML-KEM publication is exactly what you need to establish a secret backdoor. That's not popcorn, that's dynamite. I don't see how the IETF could publish the draft as is after this publication.
https://mailarchive.ietf.org/arch/msg/tls/BfV8R0Dr15PUOGfDi9d7_INJQzo/

2026-07-07 23:45:51 GMT

The MLKEM discussion in the TLS WG has clearly reached popcorn status, at least for me. As in, take the flurry of partially informed drop ins and oft repeated arguments as a source of amusement. Relax, eat popcorn or drink wine or whatever, and enjoy the show. And appreciate that someone else has to disentangle all that...

2026-06-26 20:03:58 GMT

- reply

The way I understand it, the issue is collateral fishing. Similar to fishing tuna, which is OK, but also catching dolphins, wich is not. In that case, spying on foreign agents, catching foreign data, but also catching foreign data about US citizen, which is dubious.

2026-05-08 16:48:47 GMT

- reply

I see the analogy with road networks, and the cycle of building freeways to ease traffic followed by building far away developments causing more traffic. There is always a pressure to develop more software faster, tempered by the need to fix bugs and avoid catastrophic issues. If it becomes easier to root out bugs, more software gets done faster, for an increased supply of bug. Quasi-equilibrium.

2025-12-08 20:17:50 GMT

- reply

Please be more specific. Setting a peertube instance may be hard, but the "vast majority" of users do not need to do that. They just click on a URL, and AFAIK, that works just fine with peertube.

2025-12-07 01:24:56 GMT

- reply

Hey! I use checks regularly. Companies that don't want to pay wiring fees pay me by sending a check in the mail. This looks like another case of tech bros cozying up with the goverment and making sure that everybody pays *their* fees.

2025-08-15 05:42:42 GMT

- reply

You got me to read about I2P. Your first message in this thread mentioned "inside a VPN", which I associated with what most people call a VPN, i.e., "perimeter security" solutions. I2P does not look at all like a classic VPN...

2025-08-15 05:18:08 GMT

- reply

I am sure that you are designing all that carefully. When I was at Microsoft, we developed something called "domain isolation" that used IPSEC to create a secure virtual network, so I see where you are coming from. Advantage is to reduce outside noise, and also protect legacy apps. But our IPSEC solution was later more or less abandoned in favor of just doing HTTPS from client to cloud.

2025-08-15 04:45:06 GMT

- reply

The VPN/Overlay/etc. architecture is a form of perimeter defense, of "crunchy outside, soft inside" fame. It is also quite close to the NSA smiley. I would be worried about this kind of exception. As you said, probably less worried than sending clear-text HTTP on a shared Wi-Fi connection, but still worried. Each VPN client is a point of contact with the outside world that can become a point of intrusion...

2025-08-15 00:44:52 GMT

- reply

I learned about that blog after being asked to review an Internet Draft authored my Rich Saltz, Mike Bishop and Marius Kleidl, with the mild looking title "API Keys and Privacy": https://datatracker.ietf.org/doc/draft-ietf-httpapi-privacy/. My main comment was that, hey, writing a draft is fine, but the practice of sending HTTP API credentials in clear text really should stop, and we should first expose it widely!

2025-08-15 00:33:13 GMT

Fascinating blog that reveal a worrying practice: https://jviide.iki.fi/http-redirects.
Many sites exposing a service with an HTTP API can be accessed on HTTP. Sure, the server will redirect to HTTPS, but the client has already sent credentials over the HTTP connection, basically sending password-equivalent data in clear text! Consequences are easy to guess. The blog lists dozens of servers doing that, some small, several quite big!

2025-07-31 16:35:37 GMT

- reply

I was little, maybe 7 or 8, and that day my mom arranged for me to get lunch at school instead of at home. Lunch was served in big tables by the schoolmaster's wife. "Madam, can I get a knife?" "What do you mean, don't you have your own in your pocket?"
Times have indeed changed.

2025-07-18 22:25:11 GMT

- reply

That's an old one! We came up with it for IPv6 when I was working at Microsoft. I think it was introduced in Windows XP SP1. We were worried that typing addresses like 2001:2:3::abcd was not going to work well in a variety of user interfaces, hence the "2001-2-3--abcd.ipv6-literal.net" hack. One of the team members (maybe me, not sure now) registered it on a personal account, and then transferred ownership to Microsoft before shipping.

2025-06-21 16:02:19 GMT

- reply

Nope. Link in French describes how his dogs ate the body of an 83 old man after he died.
https://www.7sur7.be/insolite/mort-depuis-plusieurs-jours-il-est-mange-par-des-chiens~ac8461cc/

2025-05-30 21:08:39 GMT

The IETF is making videos of previous meetings available by youtube. Youtube used to be freely accessible, but now requires signing in with Google account and opting into Google surveillance. Do people here believe that peertube could be a practical alternative? What would it take?

2025-03-28 17:22:15 GMT

- reply

I last did Cobol in the 70's. The modern thing then was to use "structured programming", which did improve code significantly -- but code from the 60's would not have that. And we were certainly writing tests for each program in the batch -- but who knows what lurks in the SSA code.

2025-03-28 17:11:49 GMT

- reply

it's a fascinating challenge. It is not entirely impossible, deep learning does produce translation from Chinese to English, etc. It hinges on availability of a test suite that could be used to detect errors and then reinforcement learning to improve the translator. But I wonder how they came with a two months figure. Two years, maybe. And a huge budget.

2025-03-25 04:54:02 GMT

Got some feedback on QUIC Multipath during the last IETF meeting. Design is fine, we wish we had more deployments, and in particular we wish we had more experience with congestion control when multiple paths share a bottleneck. Also, what happens if lots of multipath QUIC connections move from "active path" to "backup" at the same time? So, I wrote a little piece to describe the problem...
https://privateoctopus.com/2025/03/24/multipath-congestion-control.html

2024-12-21 21:20:29 GMT

Seriously, people are still using Chrome? And complaining that a browser funded explicitly by an ad company to ensure continuous ad revenues does not support ad blockers? I have been using Firefox for years, with the EFF "Privacy Badger" add on. And yes, Mozilla is not perfect. But the DNT kerfuffle is just noise, DNT has negative value. The DNT noise reminds me of Monthy Python's "People Front of Judea"...
https://www.youtube.com/watch?v=WboggjN_G-4

2024-12-20 06:03:49 GMT

- reply

I agree with @enoclue. Linux is pretty large, and any code that size either settles you with bugs or sets you on the constant update threadmill. If you want reliability, you want a much smaller code base.

2024-11-25 20:05:43 GMT

@npub19ua…53cv wrote a blog about why he prefers the Fediverse to Bluesky. I definitely agree, except for his example of email as a distributed network that resisted enshittification. Email *is* distributed, but Microsoft and Google control enormous shares of it, so much so that most of your emails get copied on their servers, because one of the recipients use them. Wanna bet how long it will take before the enshittification of gmail?
https://www.tbray.org/ongoing/When/202x/2024/11/15/Not-Bluesky

2024-10-17 21:04:05 GMT

- reply

By "3rd order fir", do you mean interpolation using a third degree polynomial?

2024-09-08 14:52:48 GMT

- reply

Over time the landlord has a fully paid building, paid by former renters... Whether it has AC or chargers or elevators depends on market demands and regulations.

2024-09-08 06:12:03 GMT

- reply

seems like a reasonable way to do it in a condo building, if units go with assigned spots. A bit more complicated in rentals if spots are not assigned...

2024-09-08 00:49:39 GMT

- reply

Managing an EV is much harder today if you live in an apartment building. But there is no rule saying that the building cannot add a set of plugs, maybe with some control to assign costs to users. Of course we can picture evil landlords charging through the nose for charging, but this could reasonably be tackled by a mix of competition and regulations.