To clarify, a good title would be "Loading Doom entirely from DNS records"
Neither one plays Doom over DNS nor is the first paragraph in the README correct, because DNS is only abused for storage, not for computing/processing/executing instructions:
> At some point, a reasonable person asked "DNS resolves names to IP addresses, what else can it do?" The answer, apparently, is run DOOM.
> Cloudflare will serve them globally, for free, cached at the edge, to anyone who asks. They are not a file storage system. They were not designed to be a file storage system. Nobody at the IETF was thinking about them being used as a file storage system when they wrote RFC 1035. And yet here we are.
Yeah these types of hacker stories kind of bug me. They are sort of in the same vein as "you can eat for free by going to McDonald's and eating a pint of ketchup without ordering anything" or "How I drank and showered for a year using public water fountains" . Or put another way "just because you can doesn't mean you should". Trustless societies kind of suck and forcing society to lower trust by abusing trust kind of makes things incrementally suckier ("trust" here being "it's on the honor system to not abuse DNS to serve static content").
Look, if this was a project on using DNS to replace Dropbox or something, I'd agree with you.
But the demo version of Doom just isn't that large; Cloudflare will host much larger files than that for free via Cloudflare Pages/Workers. This project is clearly meant as a fun proof of concept, not some novel way to host 3 MB for free.
I’ve heard rumors that DNS records are also sometimes used in some steganography-type communications. Great way of passing small messages in a ubiquitous and innocuous system, unlikely to be blocked or raise eyebrows by accessing.
If you consider information theory, when something has states, you can store data in any system that has multiple states, which means you can store data in any system.
The placement of coffee cups on a table can be used to encode data.
At that point, only your audience needs to know that data is there.
Airplanes and many other captive portals will allow DNS traffic, but restrict everything else. Such things can be used to get free internet in such environments. It is indeed an abuse of protocol, and future protocols are going to make life difficult for everyone to prevent such abuse.
Harder Drives is such a great watch. It's also the source of a personal epiphany over the idea that you can store data in the space between you and a reflective surface or retransmittor or whatnot.
My old late friend Dan Kaminsky famously wrote the Perl module "Ozyman DNS", which allowed you to tunnel ssh session over the DNS, thus evading certain firewalls such as those controlling access to public WiFi. Modern public WiFi setups filter the DNS too, rendering this technique moot, but I remember using "Ozyman DNS" to get WiFi access on the Caltrain and that was highly satisfying.
SDo its Friday why not ask Claude do something dumb! http://github.com/elmobp/ping-doom Doom over pure icmp you need root though the other one it spat out was doom over git
I once had this silly idea to create distributed storage of arbitrary data by exploiting a range of completely unrelated sites. Say, when you want to upload your file to the System, it may store one encrypted chunk as an image on a free image hosting site, another chunk as an encoded blog post on a random forum about farming (or in the user profile?), another chunk as a youtube video, etc. Imagine having something like hundreds or thousands of such "backends". Every chunk would be stored in 3 places for high durability of course. Free storage, hidden in plain sight :) Although, I didn't think through how to store the index reliably, and, because a moderator on a random farmers' site may delete our record(s), there needs to be a system which continously validates the integrity and reuploads the chunks.
Another silly (compression-based) idea I had was to:
- Index say google images, or something else with a large amount of URL -> data
- Find patterns in the indexed data that match patterns in your data, such that storing the URL and an offset into the data (or something more complex) would be smaller than the data chunk you are trying to compress
- Repeat for all chunks
- After you're done you can run it again and again. Infinite compression!
Yes the user has to download WAY more data that what they are trying to extract, and you'd need an insanely large index to be able to compress, but hey it was an idea.
I seem to remember bigtable also being interesting.
More than that, you might enjoy MIT’s distributed systems course. It’s all freely available online. I went through it for fun a decade ago or so, and it’s worthwhile for reasoning through hard problems like this.
People have definitely (ab)used YouTube as a filesystem though. And that’s probably your best bet for durability and performance.
>They are not a file storage system. They were not designed to be a file storage system. Nobody at the IETF was thinking about them being used as a file storage system when they wrote RFC 1035. And yet here we are.
That's plain wrong. DNS was specifically designed to store arbitrary data, cf. the Hesiod name class.
I never stop being impressed by these "<something-crazy> running Doom" posts. AFAIC, whenever we get to Mars, we won't truly have arrived until someone is playing Doom on Mars, and without wasting valuable resources by doing so. Running Doom, the canonical measurement of truly mastering a thing's capabilities.
very cool, i did something similar but turning the doom frame running on a server into ascii (with colour) and then a small shim to give inputs via subdomains
(1) A DNS file drop: Split small files into TXT records and rebuild them client-side. Useless for big files, perfect for config blobs, tiny payloads, and cursed demos. Also someone can write an S3-compatible client.
(2) Redis DNS:
- GET foo.cache.example.com -> TXT record returns value chunks
- TTL is the eviction policy
- Cache invalidation becomes even more of a hate crime.
Another fake Doom run, like the predictor one. This doesn't actually run Doom. Sorry. Meanwhile, other esoteric platforms actually runthe software.
There's the Infocom ZMachine with Zork I-III, Tristam Island, Calypso (Z machine v3 games) and many more which can be run starting from a PostScript file to a pen, a simple FPGA machine, an Amiga, the original Game Boy and who knows what.
If you can port a libre interpreter, you can run it. Old PDA's, Smartphones, JS browsers, Windows 95 machines with Winfrotz, DOS, Raspberry Pies with GNU/Linux, Riscos... There are emulators even written in Perl, Python, Lua, tons of them. It's text based output and the Z machine format it's documented.
I think some Activision games had the the Zork game embedded on their engine as an Easter Egg. As it's an 'easy' task for any programmer embedding it under a fake ingame computer woudn't have been a daunting task.
Maybe I can adapt the PostScript one to Eforth under the Subleq VM, PS' syntax maps slightly ok to EForth...
With Asterisks and some old modules you can even play it over a VOIP client and listen to the output with Flite/Festival/Espeak-nG or any compatible TTS software, such as PicoTTS. The voice input it's done with CMU Sphinx.
We've known for years that we tunnel IP over DNS [1]. We know, of course, that we can load or play DOOM over IP. Suddenly, combining the two things we already know how to do is supposed to garner attention and plaudits?
I read this title, did a double-take, then had to go look at the git hub because it still didn't click for me. Because this sounds absolutely amazing, and absurd, and weird, all at the same time. Like..... Wow? Talk about turning protocols into pretzels...
> had to go look at the git hub because it still didn't click for me
Obviously it still didn't click for you or you're lying about looking at the GitHub, because if you did, you'd have learned that it's not using DNS to run DOOM, only to store it. Which...shouldn't really be a surprise to anybody who knows that DNS TXT records exist.
And obviously your forgetting that doing this is from my perspective a very novel idea and I didn't consider a TXT record as a data storage system. Good grief.
To clarify, a good title would be "Loading Doom entirely from DNS records"
Neither one plays Doom over DNS nor is the first paragraph in the README correct, because DNS is only abused for storage, not for computing/processing/executing instructions:
> At some point, a reasonable person asked "DNS resolves names to IP addresses, what else can it do?" The answer, apparently, is run DOOM.
Yup. A better title might be “Author discovers data can be stored in DNS TXT records which were created to store data.”
Data can be stored in A records, too, just less efficiently.
(Or AAAA, or CNAME, or…)
[dead]
Also we could probably achieve this by using dnsfs and regular doom install
https://blog.benjojo.co.uk/post/dns-filesystem-true-cloud-st...
You make me wonder if it is possible. All you need to do is to programmatically change bits, and you have compute. Some cache monkeying or somethong.
Of course, I imagine it would be incredibly slow.
> All you need to do is to programmatically change bits, and you have compute.
all you need is to rapidly push off one foot and land on the other, and you have running.
> Cloudflare will serve them globally, for free, cached at the edge, to anyone who asks. They are not a file storage system. They were not designed to be a file storage system. Nobody at the IETF was thinking about them being used as a file storage system when they wrote RFC 1035. And yet here we are.
Yeah these types of hacker stories kind of bug me. They are sort of in the same vein as "you can eat for free by going to McDonald's and eating a pint of ketchup without ordering anything" or "How I drank and showered for a year using public water fountains" . Or put another way "just because you can doesn't mean you should". Trustless societies kind of suck and forcing society to lower trust by abusing trust kind of makes things incrementally suckier ("trust" here being "it's on the honor system to not abuse DNS to serve static content").
Look, if this was a project on using DNS to replace Dropbox or something, I'd agree with you.
But the demo version of Doom just isn't that large; Cloudflare will host much larger files than that for free via Cloudflare Pages/Workers. This project is clearly meant as a fun proof of concept, not some novel way to host 3 MB for free.
I’ve heard rumors that DNS records are also sometimes used in some steganography-type communications. Great way of passing small messages in a ubiquitous and innocuous system, unlikely to be blocked or raise eyebrows by accessing.
A popular use of DNS is for malware to communicate their status. They do this by requesting e.g. "i_am_in_$RANDOM_NUMBER".badplace.ru.cn.cx.
If you consider information theory, when something has states, you can store data in any system that has multiple states, which means you can store data in any system.
The placement of coffee cups on a table can be used to encode data.
At that point, only your audience needs to know that data is there.
I mean, kind of, but they're able to be cached easily and inexpensively in a way that kind of defies the intrinsic values behind steganography.
Not cache-able if no one has seen them before.
it can be used as a novel way to host files.
Airplanes and many other captive portals will allow DNS traffic, but restrict everything else. Such things can be used to get free internet in such environments. It is indeed an abuse of protocol, and future protocols are going to make life difficult for everyone to prevent such abuse.
This novel form of data storage reminds of me of this classic YouTube video, Harder Drive: https://www.youtube.com/watch?v=JcJSW7Rprio
Harder Drives is such a great watch. It's also the source of a personal epiphany over the idea that you can store data in the space between you and a reflective surface or retransmittor or whatnot.
It reminds me of the mercury delay lines - https://en.wikipedia.org/wiki/Delay-line_memory
"I have one memory system remaining. Delay-line."
https://vgel.me/fiction/gyre/
My old late friend Dan Kaminsky famously wrote the Perl module "Ozyman DNS", which allowed you to tunnel ssh session over the DNS, thus evading certain firewalls such as those controlling access to public WiFi. Modern public WiFi setups filter the DNS too, rendering this technique moot, but I remember using "Ozyman DNS" to get WiFi access on the Caltrain and that was highly satisfying.
https://boingboing.net/2004/06/21/tunneling-ssh-over-d.html
Waiting for Doom over https://github.com/yarrick/pingfs next
^ This is amazing!
Thanks for the share!
SDo its Friday why not ask Claude do something dumb! http://github.com/elmobp/ping-doom Doom over pure icmp you need root though the other one it spat out was doom over git
I once had this silly idea to create distributed storage of arbitrary data by exploiting a range of completely unrelated sites. Say, when you want to upload your file to the System, it may store one encrypted chunk as an image on a free image hosting site, another chunk as an encoded blog post on a random forum about farming (or in the user profile?), another chunk as a youtube video, etc. Imagine having something like hundreds or thousands of such "backends". Every chunk would be stored in 3 places for high durability of course. Free storage, hidden in plain sight :) Although, I didn't think through how to store the index reliably, and, because a moderator on a random farmers' site may delete our record(s), there needs to be a system which continously validates the integrity and reuploads the chunks.
Maybe such a silly project already exists?
I had the same idea!
Another silly (compression-based) idea I had was to:
- Index say google images, or something else with a large amount of URL -> data
- Find patterns in the indexed data that match patterns in your data, such that storing the URL and an offset into the data (or something more complex) would be smaller than the data chunk you are trying to compress
- Repeat for all chunks
- After you're done you can run it again and again. Infinite compression!
Yes the user has to download WAY more data that what they are trying to extract, and you'd need an insanely large index to be able to compress, but hey it was an idea.
You might enjoy reading through the original Google FS papers. I forget what they’re called but it addresses the durability problems.
Ah, I couldn’t remember the name because it’s literally named Google File System. https://static.googleusercontent.com/media/research.google.c...
I seem to remember bigtable also being interesting.
More than that, you might enjoy MIT’s distributed systems course. It’s all freely available online. I went through it for fun a decade ago or so, and it’s worthwhile for reasoning through hard problems like this.
People have definitely (ab)used YouTube as a filesystem though. And that’s probably your best bet for durability and performance.
You could write some custom backends for https://irmin.org/ I guess.
> Irmin is an OCaml library for building mergeable, branchable distributed data stores.
I've had this exact idea. Would need to be error encoded to account for chubks disappearing. There would be a rot rate as sites die or change.
lol now I wanna build this. It's like the dark web but without user or in this case, site consent. This could be a fun few weekend project
>They are not a file storage system. They were not designed to be a file storage system. Nobody at the IETF was thinking about them being used as a file storage system when they wrote RFC 1035. And yet here we are.
That's plain wrong. DNS was specifically designed to store arbitrary data, cf. the Hesiod name class.
Gotta admit that it didn't occur to me that "can it run DOOM?" would stretch all the way to DNS.
At this point I am wondering if people will somehow port DOOM over to the MONIAC.
You were right to assume that in this case. DNS is not running doom here, it's just storing it.
That's fair. I guess "can it store DOOM?" is still an interesting question though.
Is it? DNS has an explicit mechanism for storing data.
Ok well it was new to me ok!
Which is more ambitious, targeting the MONIAC platform or ENIAC?
https://en.wikipedia.org/wiki/ENIAC
https://en.wikipedia.org/wiki/Phillips_Machine (MONIAC)
I'd say both are looking increasingly doable.
“Run” is doing a lot of heavy lifting at this point.
I remember the pregnancy test Doom. Wasn't it "running" on the display only?
Yes, I think it was, but that was also b/c, IIRC, the pregnancy tester had a CPU, too. A CPU can actually run things.
DNS … cannot, and that's why the person upthread is criticizing the use of the word "run" here. DNS ran nothing.
They replaced both the display and the microcontroller with something like an Adafruir Trinket.
It wasn’t even the test’s original display, IIRC; they just re-used the housing and inserted a fancier display.
Coming up: playing doom on Ping-as-Storage
It's ready, nothing to enjoy.
ICMP packets can be arbitrary length, and there are plenty of tcp-over-icmp tunnels.
A database storing data? Now I’ve seen everything!
I never stop being impressed by these "<something-crazy> running Doom" posts. AFAIC, whenever we get to Mars, we won't truly have arrived until someone is playing Doom on Mars, and without wasting valuable resources by doing so. Running Doom, the canonical measurement of truly mastering a thing's capabilities.
I am still waiting someone to port Doom on CSS
Someone delivered that 2 years ago: https://codepen.io/cobra_winfrey/pen/oNOMRav
Oddly, it doesn't work on my browser (Firefox on Gnome).
what is the frame rate?
any of you did doom inside meta glasses?
very cool, i did something similar but turning the doom frame running on a server into ascii (with colour) and then a small shim to give inputs via subdomains
https://www.youtube.com/watch?v=GoPWuJR6Npc
without the colour i did it in a worse way for bad apple
https://www.youtube.com/watch?v=AJ2Q12vYojY
Finally, a DOOM download that bypasses captive portals
Respect. But also ... WHY????
Now let's do
(1) A DNS file drop: Split small files into TXT records and rebuild them client-side. Useless for big files, perfect for config blobs, tiny payloads, and cursed demos. Also someone can write an S3-compatible client.
(2) Redis DNS:
- GET foo.cache.example.com -> TXT record returns value chunks
- TTL is the eviction policy
- Cache invalidation becomes even more of a hate crime.
Tangent, harder drives by suckerpinch
There's a writeup here:
https://blog.rice.is/post/doom-over-dns/
This feels so unnecessary but very cool.
Malware could still use DNS records for storage and access to bootstrapped payloads correct?
Yes, but it's not a problem, any more than downloading any arbitrary text is. You'd still have to have something execute the binary.
If anything, this would be more of a way to act as a command and control server
[dead]
[dead]
What is the serialization format? Base64?
Yes. Details here:
https://github.com/resumex/doom-over-dns/tree/main/TXTRecord...
blech...too much windows. bring me the Linux version and i might care ;)
that SVG wow how?!
thanks for doing god’s work my friend.
thank you for doing god’s work my friend.
I covered another use case for this last year:
https://youtu.be/mjSdHHJySDk
Another fake Doom run, like the predictor one. This doesn't actually run Doom. Sorry. Meanwhile, other esoteric platforms actually runthe software.
There's the Infocom ZMachine with Zork I-III, Tristam Island, Calypso (Z machine v3 games) and many more which can be run starting from a PostScript file to a pen, a simple FPGA machine, an Amiga, the original Game Boy and who knows what.
If you can port a libre interpreter, you can run it. Old PDA's, Smartphones, JS browsers, Windows 95 machines with Winfrotz, DOS, Raspberry Pies with GNU/Linux, Riscos... There are emulators even written in Perl, Python, Lua, tons of them. It's text based output and the Z machine format it's documented.
I think some Activision games had the the Zork game embedded on their engine as an Easter Egg. As it's an 'easy' task for any programmer embedding it under a fake ingame computer woudn't have been a daunting task.
Maybe I can adapt the PostScript one to Eforth under the Subleq VM, PS' syntax maps slightly ok to EForth...
With Asterisks and some old modules you can even play it over a VOIP client and listen to the output with Flite/Festival/Espeak-nG or any compatible TTS software, such as PicoTTS. The voice input it's done with CMU Sphinx.
Something Doom can't do at all.
"Author discovers that DNS stores data, and that data could be DLLs."
Okay?
[dead]
[flagged]
This is so peak
We've known for years that we tunnel IP over DNS [1]. We know, of course, that we can load or play DOOM over IP. Suddenly, combining the two things we already know how to do is supposed to garner attention and plaudits?
[1] https://code.kryo.se/iodine/
Super cool. Never thought of this. Would this be useful for seeding LLMs?
This is a data storage system, so I guess yes, data is useful to train LLMs?
Why does everything get turned into an LLM discussion?
Does this LLM discussion support LLMs?
I read this title, did a double-take, then had to go look at the git hub because it still didn't click for me. Because this sounds absolutely amazing, and absurd, and weird, all at the same time. Like..... Wow? Talk about turning protocols into pretzels...
> had to go look at the git hub because it still didn't click for me
Obviously it still didn't click for you or you're lying about looking at the GitHub, because if you did, you'd have learned that it's not using DNS to run DOOM, only to store it. Which...shouldn't really be a surprise to anybody who knows that DNS TXT records exist.
And obviously your forgetting that doing this is from my perspective a very novel idea and I didn't consider a TXT record as a data storage system. Good grief.