This map shows active Steam users all over the world, and quite frequently, there is a meme post about how there are Steam users in North Korea.
Are there Steam users in North Korea playing RDR2? No. Well…, a diplomatic answer would be that, it can be reasonably estimated the answer is highly unlikely, but as I do not have access to Steam’s data, I cannot give an opinion in absolute terms.
So, what are these dots that denote gamers in North Korea then?
This map is generated using IP geolocation data, and the key difference lies in the methodology: self-reported information (used by many competitors) versus active measurement data (used by IPinfo).
Unlike most IP geolocation providers, IPinfo does not rely solely on ISP- or ASN-reported data, such as Geofeeds, which are voluntary and self-managed. These geofeeds allow network operators to claim where their IP addresses are located—but there is no verification mechanism or policy requiring that information to be truthful, accurate, or even geographically precise.
In fact, when organizations lease IP space or use BYOIP (Bring Your Own IP), they can request that the ASN report a location that may not reflect the true usage location of the IP. Many traditional providers ingest and trust these claims as-is, and even allow customers to submit manual corrections to influence the reported location of IP ranges.
This is the foundation of how much of the traditional IP geolocation industry operates—but IPinfo takes a fundamentally different approach, which we’ll cover next.
Getting those North Korean IP addresses from our competitor
Let’s take a look at the North Korean IP address from a competitor’s data that is probably being used by Steam.
SELECT competitor_data.network
FROM loc
JOIN competitor_data
ON loc.geoname_id = competitor_data.geoname_id
WHERE loc.country='KP'
We get the following CIDRs/Networks:
104.28.229.243/32
104.28.229.244/30
104.28.229.248/31
120.219.114.25/32
45.149.92.128/32
23.147.120.247/32
57.73.214.0/23
61.149.72.86/32
104.28.67.232/31
104.28.67.234/32
104.28.70.232/31
104.28.70.234/32
104.28.84.2/31
104.28.84.4/32
104.28.99.253/32
104.28.99.254/32
104.28.100.0/32
104.28.101.253/32
104.28.101.254/32
104.28.102.0/32
104.28.118.226/31
104.28.118.228/32
104.28.121.226/31
104.28.121.228/32
104.28.197.243/32
104.28.197.244/30
104.28.197.248/31
194.50.99.122/32
194.50.111.122/32
195.178.110.208/28
196.48.114.0/24
196.56.114.0/24
196.196.114.0/24
196.197.114.0/24
196.199.114.0/24
5.62.56.160/30
185.130.45.98/32
210.52.109.0/24
175.45.176.0/22
All these IP addresses are located in North Korea, based on the data provided by our competitor, whose data is likely being used by Steam there.
Enriching the IP addresses with the ASN information
I have looked up all the ASN information using our dashboard’s enrich tool. And here are the results:
Well, Cloudflare surely can’t operate in North Korea, that is for certain. Realistically, except for AS131279, all of these IP ranges are “likely” not located in North Korea.
- Datacamp provides VPN hosting infrastructure to commercial VPN companies, enabling them to report the location of their servers in a way that makes their networks appear more geographically expansive than they actually are.
- AVAST Software s.r.o runs multiple VPN companies.
- Fibergrid specializes in IP address leasing, allowing clients to rent IP space that can be announced from various geographic locations.
- Cloudflare, like many providers with a BYOIP program, allows customers to control how their IP ranges are announced and represented. This includes the ability to submit location data (via geofeeds or other means).
All of these ASNs allow their customers to publish whatever location they want in their geofeed.
As a result, when IP geolocation providers reference these geofeeds without validation—as most do, except IPinfo—they’re essentially just repeating whatever claims those customers make. This undermines the reliability of geolocation data and creates confusion, especially in cases where accuracy truly matters.
It’s a self-reporting system with no enforcement of truthfulness, which defeats the very purpose of geolocation: knowing where things actually are, not where someone says they are.
Is our data always the superior data? Let’s admit the reality.
Now, I have mapped our location data for these IP addresses.
Unfortunately, you will see that we probably mislocated some of these IP addresses.
Our location data is based on active measurements, meaning that we actively try to identify the location of IP addresses through network diagnostic measurements that we run through about 1,000 servers. Based on the network data and our models, we can present evidence of where a particular IP address is located. This is fundamentally a superior methodology of IP geolocation than our competitors ’ methodology.
However, we actually do ingest the self-reported information presented by ASN and do consider corrections from people as well. These are considered as “fallback values”. In the edge case where we do not have the active network measurement of an IP address (maybe these devices are offline, or not assigned devices) in the first place, we have to report something. So, the only “evidence” of the location of the IP address we have is self-reported by ASNs. It is a compromise for sure, but there is no other way. We can not assign random locations to IP addresses just because this looks bad.
Take this Orange IP address range as an example, which is an inactive and undeclared IP range that Orange says is in North Korea: Why is this Orange.com IP range in North Korea?
North Korea Network Information - IPinfo.io
We could investigate IP addresses in Antarctica, North Korea, or random uninhabited islands
and assign countries manually there, but that is not how we like to do things. For us, we believe in the process that fundamentally and comprehensively represents our IP geolocation process. We will look into our whole system and evaluate what could be done on a process level to improve our data.
So, our data is undoubtedly superior to others (in this case, miles ahead). Still, systematic shortcomings cannot be addressed by any method besides the explanation of our processes of how we provide IP geolocation.
Ultimately, what we provide fundamentally and universally is a better and truthful system for IP geolocation; it is getting better every day.