I'm pretty sure it's just querying steam's server masterlist
#What determines how quickly a server pops up in server picker?
hot oyster
I'm pretty sure it's just querying steam's server masterlist
(which, to be perfectly clear, I haven't checked but that's the general way to implement this stuff)
pale field
Yeah it would make the most sense but this behaviour can't be replicated in steam server browser
hot oyster
of course you can filter servers in order of ping but I don't think that has relevance to the order of them showing up
hot oyster
Yeah it would make the most sense but this behaviour can't be replicated in stea...
oh that's interesting
pale field
I'm asking because i've noticed that different server hosts affect how fast the server shows up
For example OVH vs Hetzner
hot oyster
Wonder what the results would be if you directly queried it through the API, pretty sure you can just do that
pale field
But the ping is similar
hot oyster
if anything I'd imagine it's proximity to Valve's own servers
pale field
That would make sense
hot oyster
but that doesn't explain server list in-game and on steam being different
pale field
I have over 2 years of data on my networks player traffic and I can directly correlate the times when I switched hosts
hot oyster
I guess the best way to test would be getting two servers as physically close to each other as possible and seeing if there's still a difference
I could see country or even city affecting things a lot
pale field
Red line is when I switched hosts
The ping for my players is better than it was
But it clearly affects server position in browser
hot oyster
just for reference did you change physical location, particularly country, and from what to what did you switch?
hot oyster
it's also interesting to see seemingly lower average but higher variance?
pale field
Where my players are
pale field
it's also interesting to see seemingly lower average but higher variance?
It's a mystery for me too
hot oyster
I'd imagine the spikes would be people from western EU or US?
pale field
Player demographics didn't change
hot oyster
particularly US would suffer a lot from the change
hot oyster
Player demographics didn't change
I mean that doesn't mean it couldn't be explained with that I think
pale field
I have detailed monitoring on that too
hot oyster
higher spikes could IMO be explained by a minority having a much worse ping
but also maybe not
hot oyster
could realistically also be that
as a side note how do you monitor that?
I mean I know how it could be done with IP tracking since those are geolocated but I'm interested in the specific solution, particularly if it's public
pale field
We screen all IPs connecting to our services for VPNs and steam accounts in bad standing
hot oyster
is that a custom solution or something public?
pale field
Everything on my network is custom I'm a dev
hot oyster
I wasn't actually aware of there being a way to know account standing based on IPs
hot oyster
Everything on my network is custom I'm a dev
I mean I assumed it's at least partially made by you
pale field
Cheats usually come in a bundle with HWID spoofer and a proxy in India that we detected some time ago
We block certain ranges of IPs
pale field
I wasn't actually aware of there being a way to know account standing based on I...
Oh thats from steam api
hot oyster
ah, that makes much more sense
pale field
Account age, level, and bans etc
hot oyster
I knew you could get that info there :p
pale field
We calculate a risk score based on that
hot oyster
I think you've talked about that earlier too
hot oyster
just in this discord in general
things of this nature are of interest to me so I tend to remember at least bits and pieces - not that I can make something like this myself just yet (but maybe in a couple years...)
pale field
Yeah I'm much more into making cool infrastructure and tech stuff than in to running my network xd
hot oyster
Yeah as am I
pale field
We are also experimenting with AI now
hot oyster
currently looking into making an ELO system (Glicko-2 specifically) based on just server logs (or possibly just going to use the ptero API to grab the data directly from console) because why not
seemed like a fun idea until I actually looked into Glicko-2
pale field
We have something similar but it's not public yet
hot oyster
We are also experimenting with AI now
I'll probably be starting on that in a few years after I get my bachelor's (data science)
realistically the only reason I want to make such a system is to see how the data looks like, particularly as it absolutely isn't designed for something like this