←back to thread

Cross-Platform P2P Wi-Fi: How the EU Killed AWDL

(www.ditto.com)
223 points stusmall | 4 comments | 28 Mar 25 13:13 UTC | HN request time: 0.628s | source
Show context
apitman ◴[28 Mar 25 23:34 UTC] No.43511040[source]▶
I frequently play Age of Empires 2 with my wife and her brother. Remarkably, this game still supports LAN play in 2025, even though the netcode has been completely overhauled since 1999.

However, we decided to try it on a recent flight, and it turns out it still requires an internet connection, both to satisfy Steam, and to connect to some sort of LAN coordination server. I ended up paying $20 for in-flight wifi.

We've lost a lot in the last 30 years, but tech like wifi aware might help bring back local-first networking. I choose to believe that if solid APIs exist, developers will use them.

replies(1): >>43511290 #
1. dathinab ◴[29 Mar 25 00:04 UTC] No.43511290[source]▶
> and to connect to some sort of LAN

this is actually kind of a hard UI/UX problem for game developers

many p2p+local auto recovery protocols are very bothersome, partially due to some of the protocols being bad or incomplete and a lot due to all kind of hardware & OSs partially or fully crippling them

so game devs often have to fall back to a coordinator server to provide reliable and easy to use functionality for most which also happens to often be the easiest thing to implement and maintain, and then in addition they could also implement work-arounds for the no-internet case

but that is additional cost for a overall niche use case (local co-op without internet), so it ends up in the backlog with low priority at best or gets outright killed. To make that worse steam provides tools to make it much easier to implement co-op (focused on non local co-op), and the easiest way to use them is in a way which always requires internet even for local co-op

so as long as steam doesn't put in a lot of work to make no-internet local co-op close to free to implement for most games it will never happen for most games

replies(2): >>43511341 #>>43513144 #
2. apitman ◴[29 Mar 25 00:08 UTC] No.43511341[source]▶
All good points. This is why I'm so excited about this development. Maybe wifi aware will provide a reliable means of local discovery. At that point we're just missing an open source library that makes the devex for implementing LAN support as good as Steam's, and baby you got a stew goin.
3. fulafel ◴[29 Mar 25 05:59 UTC] No.43513144[source]▶
What's the problem with just sending some UDP beacons to the local network broadcast address? (or multicast all-hosts address)
replies(1): >>43528452 #
4. dathinab ◴[30 Mar 25 22:41 UTC] No.43528452[source]▶
- middleware in the networks not forwarding broadcast messages

- depending on device and application type you not even being able to send broadcast/the OS silently dropping them

- firewall blocking incoming TCP/UDP without hole punching

- p2p in games having security implications (unsafe network stacks, game engine etc. allowing RCEs and similar) so you want to make sure only "more trusted" communication can happen, so TLS is needed, but without actually fully secure p2p TLS is not easy, mainly there are issues with establishing trust (you either have to involve some side channel (i.e. a pin, QR code or similar) or pre-established trust.

The biggest thing is still that as a steam game you have a reliably, proven, easy to use "solution" as part of your normal steam integration which you anyway want to use to be able to use the friend invite system which has the drawback of needing internet for local coop which is niche use-case likely not selling any games. Why would a company implement an additional solution and handle all the UX issues of switching between them?