Mastodawn

Concerning, bc it indicates that the extent of the concentration of power in the hands of a few hyperscalers is way less widely understood than I’d assumed. Which bodes poorly for our ability to craft reality-based strategies capable of contesting this concentration & solving the real problem. 2/

The question isn’t "why does Signal use AWS?" It’s to look at the infrastructural requirements of any global, real-time, mass comms platform and ask how it is that we got to a place where there’s no realistic alternative to AWS and the other hyperscalers. 3/

Running a low-latency platform for instant comms capable of carrying millions of concurrent audio/video calls requires a pre-built, planet-spanning network of compute, storage and edge presence that requires constant maintenance, significant electricity and persistent attention and monitoring. 4/

Instant messaging demands near-zero latency. Voice and video in particular require complex global signaling & regional relays to manage jitter and packet loss. These are things that AWS, Azure, and GCP provide at global scale that, practically speaking, others (in the western context) don’t. 5/

This isn't ‘'renting a server.' It's leasing access to a whole sprawling, capital-intensive, technically-capable system that must be just as available in Cairo as in Capetown, just as functional in Bangkok as Berlin. Particularly given the high stakes use cases of many who rely on Signal. 6/

Such infrastructure costs billions and billions of dollars to provision and maintain, and it’s highly depreciable. In the case of the hyperscalers, the staggering cost is cross-subsidized by other businesses–themselves also massive platforms with significant lockin. 7/

Meaning that infrastructure like AWS is not something that Signal, or almost anyone else, could afford to just “spin up.” Which is why nearly everyone that manages a real-time service–from Signal, to X, to Palantir, to Mastodon–rely at least in part on services provisioned by these companies. 8/

But even if Signal had the billions needed to recreate AWS, it’s not just about money. The talent to run these systems is rare & concentrated. The expertise, the tooling, the playbooks, the very language of modern SRE came out of these hyperscalers, and is now synonymous with 'the cloud.' 9/

o, yes, Signal runs on AWS. It also runs on your phone, which runs on iOS (Apple) or Android (Google). And on Dekstop, via Windows (Microsoft). Each of these presents similar dependencies on large entrenched tech companies, and concomitant barriers and risks. 10/

Meredith Whittaker

In short, the problem here is not that Signal ‘chose’ to run on AWS. The problem is the concentration of power in the infrastructure space that means there isn’t really another choice: the entire stack, practically speaking, is owned by 3-4 players. 11/

So, Signal does what we can to provide a service w integrity in the concentrated ecosystem we're working in. We protect your comms w end-to-end encryption, so that we can use AWS and others as a highway across which to send Signal data in ways that don’t let AWS, or anyone else, gain access. 12/

To conclude: my silver lining hope is that AWS going down can be a learning moment, in which the risks of concentrating the nervous system of our world in the hands of a few players become very clear. And that this can help us craft ways of undoing this concentration and creating real choice ❤️ 13/

yawnbox🔜fosdem Oct 27

@Mer__edith

The tor network has had 100% uptime. 100%

@yawnbox @Mer__edith Try running video calls over Tor.

yawnbox🔜fosdem Oct 27

@davep

i did not say run Signal over tor

@yawnbox So what's your point?

Laurens 🧢 🐐 🔻Oct 27

@davep @yawnbox Don't feed the troll.

iwein Oct 27

Decentralisation is a valid strategy?
@yawnbox

David Fetter Oct 27

@iwein Did you read literally any part of the thread? Some people will choose low reliability, high latency, and low availability so they can attempt decentralization. Some people will also become hermits. It's about the same proportion of people.

iwein Oct 27

@davidfetter I did read the whole thing, and this was the only point I could distill from the tor analogy. You're not wrong that it's not going to be comfortable. I was happy to explain it to the one that brought it up, if that was indeed the point being made.

I don't have a strong opinion here, happy signal user, not trying to add hot takes of my own 🙂🍿

Sorry if that was unclear.

FinchHaven sfba Oct 27

@yawnbox

"i did not say run Signal over tor"

Run Tor over Tor

Got it

Insightful

cc @davep

@FinchHaven @yawnbox @davep except @signalapp chose to use #AWS (and @Mer__edith chose to not migrate away from it) as well as #Signal chose to bloat from #TextMessaging to centralized #VideoCalling.

There are better options and even if Signal refused to #decentraloze they could've easily outsorced the most workload to users with #P2P #WebRTC.

If people accept nonexisting dedicaded Servers for AAA FPS games sold for $60+, then they certainly can deal with P2P videocalls in a "Free"* App…
Free as in: Relying on big donations because the people behind it can't even do a #Shitcoin #Scam properly...

Kevin Karhan :verified: (@kkarhan@infosec.space)

@jenzi@mastodon.social @lexinova @davidfetter@kolektiva.social Doesn't even need that. #XMPP+#OMEMO (Chats) & #P2P-#WebRTC (Video- & VoiceCalling) are *evdently better* than a centralized, #SingleVendor & #SingleProvider [system](https://infosec.space/@kkarhan/114935952643402592)… - And that is why @signalapp@mastodon.world will inevitably end like #ICQ & #Skype!

Infosec.Space

@davep You can run video calls over jitsi.

IPv6 was supposed to solve NAT, so fewer servers would be needed. Supposed to.

@yawnbox

@ArneBab @davep @yawnbox This is 2025, and there are still large ISP's with no intention to offer IPv6.

@wanwizard And there are ISPs with IPv6 whose addresses cannot be accessed from outside the ISP.

That’s why we cancelled a fiber plan here: we found out that half the people wouldn’t be able to connect via IPv6, and there was no IPv4 at all.

It seems that they fail badly on their peering.
@davep @yawnbox

@ArneBab @wanwizard @yawnbox That's bonkers!

@davep @ArneBab @yawnbox Yup. In the UK, if you want an ISP that can give you unfiltered, fixed IPv4 and IPv6, it's slim pickings. You are limited to a few small providers, or very expensive business packages.

@wanwizard I don’t even need fixed IPv4. I have a dyndns address.

But I need a *reachable* address.
@davep @yawnbox

@ArneBab @wanwizard @yawnbox I guess we're lucky here in France. We can get a fixed IPV4 address at no extra cost.

@davep @ArneBab @yawnbox For home connections, only a few offer fixed IPv4 at all (none of the large national providers do), and even fewer for free.

@ArneBab @davep @yawnbox Sounds like the same problem as you have with mobile providers. CGNAT.

@wanwizard @ArneBab @yawnbox We have that on our backup 4G network, but that's normal really.

@davep it may be normal, but it shouldn’t be.

We have this awesome distributed structure with full end-to-end connections between all devices and we use it to build a network where devices cannot talk to each other …

… and then we wonder why AWS became part of the failure point trinity of modern computing.
@wanwizard @yawnbox

@davep @ArneBab @yawnbox Normal as in "we've gotten used to it", yes.

But definitely not normal. It is technically not an issue to give every mobile connection a publicly reachable IPv6 address, it is what it is designed for.

CGNAT is one of the things that makes both ISPs and hosting parties lazy, there is still plenty out there that is IPv4 only.

I often see surprised faces when I discuss our hosting services with potential clients, and I mention that we run full dual stack everywhere.

@wanwizard @davep @ArneBab @yawnbox IMHO #CGNAT should've been outlawed and #IPv6 support should've been mandated by law...

But that's beyond the point…

Kevin Karhan :verified: (@kkarhan@infosec.space)

@wanwizard@mastodon.scot @davep@infosec.exchange @ArneBab@rollenspiel.social @yawnbox@disobey.net even my #ISP won't give me proper #DualStack, so I'm stuck on static #IPv4 only… https://infosec.space/@kkarhan/115492154062048948

Infosec.Space

https://infosec.space/@kkarhan/115492154062048948

@wanwizard @davep @ArneBab @yawnbox even my #ISP won't give me proper #DualStack, so I'm stuck on static #IPv4 only…

Kevin Karhan :verified: (@kkarhan@infosec.space)

@FinchHaven@sfba.social @yawnbox@disobey.net @davep@infosec.exchange except @signalapp@mastodon.world *chose* to use #AWS (and @Mer__edith@mastodon.world chose to not migrate away from it) as well as #Signal *chose* to bloat from #TextMessaging to centralized #VideoCalling. There are [better options](https://infosec.space/@kkarhan/115492122419368937) and even if Signal refused to #decentraloze they could've easily outsorced the most workload to users with #P2P #WebRTC. - If people accept *nonexisting dedicaded Servers* for [AAA FPS games sold for $60+](https://www.youtube.com/watch?v=tBvFmhfllrk&t=4m48s), then they certainly can deal with P2P videocalls in a *"Free*"* App… * Free as in: Relying on big donations because the people behind it can't even do a #Shitcoin #Scam [properly...](https://www.youtube.com/watch?v=0DSGq9FQKU4)

Infosec.Space

WanWizard 🏴󠁧󠁢󠁳󠁣󠁴󠁿 🇳🇱Nov 4

@kkarhan @davep @ArneBab @yawnbox At least it is static, most providers here don't even give you that.

Francis 🏴‍☠️ Gulotta Oct 27

@ArneBab @davep I know that it's not a conspiracy that we don't use IPv6 very much, but I also feel in my bones that it's a conspiracy that we don't use IPv6 very much.

Preston Maness ☭Oct 28

@reconbot @ArneBab @davep it need not be a traditional conspiracy when the big players all share a common interest in maintaining the dominant broadcasting^W client-server model.

Billiglarper Oct 27

@ArneBab @davep @yawnbox

Right now, Jitsi doesn't run video calls reliably. Even if it's just a handful of users, all within a few km of Frankfurt. (And we are not even using it for video conferencing. Most are voice only, perhaps 1-2 persons sharing a screen.)

@billiglarper ouch, that’s painful. I’ve done Jitsi calls with 30 people.

Did that degrade with updates or was it always that bad for you?

@billiglarper An alternative would be BBB. We’ve been using that for 4-person roleplaying sessions for years now. All with Video.

Billiglarper Oct 27

@ArneBab @davep @yawnbox

What's the point of this discussion? It's local anecdotal evidence in a thread on large global infrastructure issues by an industry expert.

(Yes, Jitsi was always janky. But it's not a messenger, and it doesn't even do end to end encryption in most cases.)

@billiglarper trying to check for points where centralization can be reduced, because the AWS outage showed how dependent Signal is on centralized infrastructure.

Note that Meredith is not in the recipients to avoid filling her replies with the unresolved discussion.

DrYak Oct 27

@ArneBab @davep @yawnbox Note that in the specific use case of Signal: given their threat model, "direct peer-to-peer connections by default" are not desirable. You'll need to bounce the audio&video traffic by default to make it more costly to infere who is talking with whom.

So the fact that working NAT and IPv6 help rely less on TURN servers won't help decentralize that much.

@dryak Maybe a start could be to switch to direct peer-to-peer connection if Signal sees that both sides are in the same subnet (i.e. on the same wifi).

In that case they connect to the signal server for connection with a voice-data profile *at the same time* which already gives away that they are talking, so staying in the subnet with a direct peer-to-peer connection would reduce the total privacy loss.

@dryak but firstoff, to stop discussing with too little information: the reason they use a forwarding server is that a single device can’t send video to 40 people via direct connections.

Here’s their description: https://signal.org/blog/how-to-build-encrypted-group-calls/

That also shows the level of complexity involved already.

How to build large-scale end-to-end encrypted group video calls

Signal released end-to-end encrypted group calls a year ago, and since then we’ve scaled from support for 5 participants all the way to 40. There is no off the shelf software that would allow us to support calls of that size while ensuring that all communication is end-to-end encrypted, so we bui...

Signal Messenger

https://github.com/signalapp/libsignal?tab=readme-ov-file

@ArneBab @dryak
A mutual friend sent me Moxie's original white paper for Signal's use of TPMs, but I lost it and can't find reference to its use now (it's been nearly a decade...). Anyone got any resources? It's potentially an impediment to decentralisation, but my memory is hazy.

Things may have moved on since then, "attest: Functionality for remote attestation of SGX enclaves and server-side HSMs."

@davep @ArneBab @dryak yeah, the same #proprietary shitboxes thar get hacked so often.that #Intelcyeeted that from #Consumer #CPU|s and now there's no "legal" way to play #4K #BluRayDisc|s on modern systems.

Moxie trusts too much into the silicon of parties who's goals are irreconcileable at odds with his demands.

But I guess that's normal with @signalapp folks…
https://infosec.space/@kkarhan/114935952643402592
https://infosec.space/@kkarhan/115492122419368937
https://infosec.space/@kkarhan/115492154062048948

Kevin Karhan :verified: (@kkarhan@infosec.space)

My [reservations](https://infosec.space/@kkarhan/114234551915193036) and [criticism](https://infosec.space/@kkarhan/114862595629371002) re: #Signal are not just valid, but the reality is *even worse than I thought*: - The fact that @signalapp@mastodon.world requires not only their shitty #Android #App, and a #PhoneNumber but literally won't allow people to use their shitty #Desktop-App unless they have an Android device with a camera pointed at it makes it utterly unuseable for certain users *who don't have a fucking #camera in their Android*… Seriously, do they expect folks to deal with that shit? - It's already worse in terms of #UX than #telegram and #discord and that too makes #XMPP+#OMEMO clients like @monocles@monocles.social / #monoclesChat & @gajim@fosstodon.org / #gajim easier and faster to onboard #TechIlliterates onto. - Whichever asshole decided that a *replacement for #SMS* should mandate #PII like a #PhoneNumber & not be natively cross-platform should be banned from doing any #tech in their life. Trying to circumvent this shit and helping folks with it makes me so fucking angry that I'm now explicitly refusing to support it! FIX THAT SHIT, @Mer__edith@mastodon.world, and if it means you need to kick some devs in their crouch then consider this a necessary *"investment"*… #sarcasm #TechSupport #TapesFromTechSupport #Enshittifucation #SignalSucks #TelegramSucks #Messengers

Infosec.Space

zaire arcana Oct 27

@davep @yawnbox @Mer__edith Bandwidth would not be an issue I’m fairly certain.

@soop @yawnbox @Mer__edith Latency is the issue, not bandwidth.

@davep @soop @yawnbox @Mer__edith no it's not.

It's a solved problem even for AAA FPSes sold for $60 - 7 years ago!

Kevin Karhan :verified: (@kkarhan@infosec.space)

Infosec.Space

Max Lee Oct 27

@davep @yawnbox @Mer__edith Pretty sure that wouldn't even be as big of an issue as long as you don't try to exit the network.

You could even potentially improve the throughput ability by making every client that wants to use the network a node that relays traffic when it doesn't have active calls, however that's not suited to be automatically activated on mobile devices with limited power or even data caps. (But I would imagine people would be willingly donate resources to such a network if a simple separate application was offered the same way as it's done with TOR already)

DrYak Oct 27

@davep @yawnbox @Mer__edith Regarding Tor: instant messaging (if you stretch "instant" to cover several seconds which is acceptable in practice) have been successfully ran over Tor and other distributed settings.

Regarding video not relying on a centralized infra: Skype during its Kazaa-/pre-Microsoft- era and its "Super nodes" has been a widely successful example of a video calling software that doesn't rely that much on centralisation (but of course with a completely different security model)

DrYak Oct 27

@davep @yawnbox @Mer__edith (note: I am not saying that Signal is bad. Merely jumping in about centralisation. I actually appreciate Signal, e.g., unlike the above example of Skype, it is tolerating 3rd party open source clients, so people like me who neither run Android nor iOS on the smartphone can still communicate with friends).

@davep @yawnbox @Mer__edith

Diogo Constantino Oct 27

@dryak also skype was crap that failed every day for many of us.

Preston Maness ☭Oct 28

@dryak @davep @yawnbox @Mer__edith I've run my Signal clients (mobile, desktop) over tor for years. For chat at least, it's rarely noticable. Calls and video struggle though.

Oct 28

@aspensmonster @dryak @yawnbox @Mer__edith
That seems feasible enough. But there's a big difference between having a decent enough experience using Tor for a subset of client interactions and using it across the Signal infrastructure.

I guess the benefit would be to make it (somewhat) more difficult for third parties to link people using synchronous communications, but I don't know if that's a big enough incentive to try it (and it's the synchronous comms that do worst with Tor).

I'm no expert and just thinking out loud here 🤪

@dryak @davep @yawnbox @Mer__edith yes, cuz.I've been using #XMPP (#OTR, now #OMEMO) for 15+ years over @torproject / #Tor on a THROTTLED #EDGEland* connection!

It's just that @signalapp chose #centralization over #sustainability…

*#Germany is EDGEland

Kevin Karhan :verified: (@kkarhan@infosec.space)

Infosec.Space

helly ✅Jan 4

@davep @yawnbox @Mer__edith theres enough bandwidth on tor for this just not with webrtc

@yawnbox I don't think you have a clear understanding of what you're talking about, and it might be fun for you to look a bit more deeply into how TOR works and its dependencies.

yawnbox🔜fosdem Oct 27

@Mer__edith

With respect Meredith, i’m talking about decentralized protocols and their capability to not depend so heavily on the service providers you’re arguing for. Tor Project has shown how possible it is (i used to work there, and it’s spelled Tor not TOR).

I listened to Moxie’s aversions to decentralization for years. That’s what I keep seeing now, with posts like these. I also understand the value of huge cloud providers, I’ve worked for many companies who use them, and have worked for them, and I understand why you depend on them and how important that is to a high quality service. Thank you for all that you all do.

But what conversations does Signal Foundation actually have on the topics of resiliency through decentralization? How much money could you save by allowing the community to take on aspects of the network? How much resiliency and trust could be gained, without losing performance?

Tris Oct 27

@yawnbox @Mer__edith There's @cwtch which uses Tor for routing. All good except... it's P2P xD

And I did try to like Signal - honestly!

@tris @yawnbox @Mer__edith @cwtch granted, @signalapp chose to be bad!

As in: It's easer, faster, cheaper, more resilient, private and secure to onboard #TechIlliterates woth #XMPP+#OMEMO over @torproject / #Tor using @guardianproject #Orbot @monocles / #monoclesChat than to do so for #Signal for the last 15+ years !

Kevin Karhan :verified: (@kkarhan@infosec.space)

@dryak@mstdn.science @davep@infosec.exchange @yawnbox@disobey.net @Mer__edith@mastodon.world yes, cuz.I've been using #XMPP (#OTR, now #OMEMO) for 15+ years over @torproject@mastodon.social / #Tor on a THROTTLED #EDGEland* connection! - It's just that @signalapp@mastodon.world *[chose](https://infosec.space/@kkarhan/114935952643402592) #centralization over #sustainability*… *#Germany is EDGEland

Infosec.Space

@yawnbox @Mer__edith Tor is basically a glorified network protocol (albeit very smart) so having it distributed by design is less of an issue.

I agree that making Signal more robust through decentralisation would be great, but this sort of thing gets more difficult the higher up the stack you go, especially when it wasn't part of the core design principles.

mapto Oct 28

@davep @yawnbox @Mer__edith with decentralization you're throwing out of the window not only latency, but often also the capacity to guarantee delivery.

Consider this talk for details:
https://hapyyr.com/@bogo/115401249466782443
...also for alternatives to Signal and the corresponding tradeoffs.

Bogomil Shopov - Бого (@bogo@hapyyr.com)

If you are looking for good privacy respecting messaging and other apps, you could check my slides from @linuxdays@mastodonczech.cz conference here : https://pretalx.linuxdays.cz/media/linuxdays-2025/submissions/KGHCKC/resources/FOSS_tools_to_fight__EE3GIn4.pdf #foss #privacy #linuxdays

Be Hapyyr - An awesome Fediverse Server

@mapto @davep @yawnbox @Mer__edith that is dangerous #disinfo, cuz you can let clients send back sending confirmation.

Worked for me for 15+ years!

Kevin Karhan :verified: (@kkarhan@infosec.space)

@tris@chaos.social @yawnbox@disobey.net @Mer__edith@mastodon.world @cwtch@fosstodon.org granted, @signalapp@mastodon.world chose to be bad! - As in: It's easer, faster, cheaper, more resilient, private and secure to onboard #TechIlliterates woth #XMPP+#OMEMO over @torproject@mastodon.social / #Tor using @guardianproject@librem.one #Orbot @monocles@monocles.social / #monoclesChat than to do so for #Signal for the last [15+ years](https://infosec.space/@kkarhan/115492199979302447) ! And [I did try to like Signal - honestly!](https://infosec.space/@kkarhan/114935952643402592)

Infosec.Space

@kkarhan @mapto @yawnbox @Mer__edith

"It's easer, faster, cheaper, more resilient, private and secure to onboard #TechIlliterates woth #XMPP+#OMEMO over @torproject / #Tor using @guardianproject #Orbot @monocles / #monoclesChat..."

If it's easier, why isn't it as successful and used by the military etc? You seem to have dismissed video calls etc too. You could argue against that functionality for certain use cases, but it's become a core part of secure messaging over time. Maybe they just have different audiences.

I kind of agree on Moxie's reliance on Intel SGX though. Even at its inception it worried me. But it's not part of the core E2EE protocol so could potentially be replaced.