Long-term investigation of unstable input, affected by electrical and network conditions

[fa1kk]

Hi everyone.
I have been facing the issue for quite a while (since ~2010) but never really documented everything I've done, which, by the way, is a lot. This is a thread about my story, the documented information, and some observations. I know that most of us would rather enjoy our time rather than desperately search for solutions. If you want to avoid wasting time like I did, consider this a quick read and feel free to point out any gaps you may notice.
Over the last 3 years I've tested hardware, power configurations, and electrical changes more than most people would ever attempt. I have worked alongside some of my boys who are experiencing similar challenges. (We are a closed group of people whom I gathered to find and fix the issue). But for the most part I will tell my perspective, sometimes including theirs. Some changes clearly affect the behavior of my PC and monitor. Others do absolutely nothing. At this point I'm no longer trying random fixes. I'm trying to understand the physical mechanism behind the problem. I want to permanently eliminate the cause so I can return to e-sports, which is something I have been doing for quite a while. I feel like the job's not done when you are not a tier 1 while being fully capable of being there.

So we start with some history of the mess. I live in Ukralne, in a city called Dnipro, in a private house on the outskirts of the city, which is literally 100 meters from me. I moved into this house in 2007, which my father built. At the time I never paid attention to the electrical installation. The house has a TN-C supply, but there is only PEN and one phase entering the house. There is no PE conductor anywhere in the house. The distribution transformer was originally from 1963 200 KVA. In October 2025 DTEK (my electric company) replaced it with a new 400 kVA transformer. The underground cable feeding the street was also repaired after multiple insulation faults were found. The entire 10 kV line from the 35 kV transformer substation was completely replaced, including all poles, fasteners, and cable, roughly 700-800 meters, and the cross-section was increased. Despite these upgrades, the core problem remained.

For those who think I’m doing this for fun, it’s not about me; I’ve been a competitive player forever. I started playing competitively around 2010, starting with WC3 when I was just a kid in elementary school. My first real love of competition was, and probably should always have been, Dota 2, until I got injuries from playing too much, which led me to quit, which is still sad but it is what it is. PUBG since 2017 was my second main game, because it has a lot of decision-making and strong shooting mechanics. Unfortunately PUBG is dead in esports, so I was forced to switch again. This time it was a tac FPS, Valorant. I've competed against high-level players, played tournaments and scrims, and reached relatively high ranks in all games I played(except valorant). I played in some orgs,including some of you probably never heard of. It doesn't matter, but it was never enough, I was always close to qualifying for bigger tournaments, sometimes I did, sometimes I even won smaller ones. I'm mentioning this only to avoid the assumption that I’m just inexperienced or blaming hardware for bad performance.

Until the end of 2023, I genuinely believed the problem was psychological and a massive skill issue. I thought I simply wasn't playing well enough. Sometimes I could outperform teammates even with the issue, just by putting in 10x more effort than most players would consider normal, just to keep up with people who had less effort and time invested. I thought it was a learning rate problem, but in reality it was outside factors. And you probably know the feeling: when learning a mechanic without the issue it takes minutes, with the issue it takes hours, which is extremely time-wasting.

After the war that began in 2022, I had a lot of electricity problems due to infrastructure damage, which led to real degradation of the transformer and the feeding line, making gaming worse to the point where a 360Hz monitor felt like 15Hz with heavy blur. In less than 3 months I tried most of the existing "tweaks," and it gave me a different perspective on Windows. Drivers and overclocking matter, but in terms of stability they are not the main issue most people deal with. I made a post on Blur Busters about the issue that kept coming back, and I got lucky because someone added me on Telegram. Later we had a Discord call where that guy told me it’s not Windows, BIOS, or peripherals, but electrical input lag. At first, I thought they were trying to lure me into some kind of a cult, because it sounded like nonsense, but the more I tested, the harder it became to dismiss.

The topic I was talking about:viewtopic.php?t=12535

For most people who already deal with this, you can skip this part, but for reference, THE SYMPTOMS:
Mouse behaviour: mouse feels heavy and viscous, small corrections become difficult, cursor feels delayed even though polling rate is correct.
Display behaviour: motion clarity changes significantly. During “good days” the monitor becomes noticeably brighter, text becomes sharper, motion blur decreases, colors become more saturated, and camera movement feels smoother.
On bad days everything feels like a replay instead of real-time gameplay. Everything is delayed and not sharp. Even typing feels delayed, and continuous mouse movement can cause nausea after a few minutes. It also happens in offline games like Kovaak’s, so it is not network related. Sometimes within one session it can switch between multiple states in minutes.
In online games everything above remains, plus additional desync-like behavior. Opponents appear to teleport or move too fast. Fast peeks become almost impossible to react to (under ~200ms). The game often feels like it is fast-forwarding multiple actions. Hit registration becomes inconsistent even on clearly landed shots.

Things that objectively change behavior (not necessarily for the better, but they change something): cable routing inside the case, different houses and areas, ethernet cable position, PSU mounting pressure, GPU screw tension, electrical work by the utility company, transformer replacement, different internet providers, routing changes, generators, battery power stations.
Things that did NOT solve the problem: multiple complete PCs, multiple laptops, different GPUs, RAM kits, monitors, mice, wired/wireless setups, different operating systems, grounding systems, isolated rods, separate circuits, UPS (all types), solar battery stations, inverters, isolation transformers, EMI filters, ferrites, ethernet isolation, Windows optimization, BIOS optimization, CPU/GPU/RAM/NIC tuning, full house wiring changes, meter replacements, cable replacements, breakers, and more. None permanently solved the issue.

And many more. I just don’t think there is a reason for another 7 pages of text. And none of them solved the issue. Including all sorts of EMI filters alongside powerful production up to 18 GHz and any kind of devices between or inside the PC, monitor, room, or peripheral.

Electrical findings: official measurements by DTEK showed excessive flicker, severe voltage unbalance, U0 outside limits, U2 outside limits, multiple voltage sags, full interruptions. Transformer-side measurements were significantly better than measurements closer to my house. I can attach documents from all three phases over multiple seasons (spring 2025 to spring 2026, yes, entire year). But they are in Ukralnlan.
The utility is currently monitoring the line 24/7 and plans to transfer ownership of the line and rebuild it, since parts are not officially owned. This process may fix it, but it is not guaranteed.
There is no difference between day and night and no difference between wireless and wired devices. The random improvement appears randomly and disappears randomly for a random amount of time; it can be minutes, hours, or sometimes even days, and there is no correlation with anything but the entire circuit of the power source.
I also built a personal grounding system: 7 rods, one 10 meters and others 2 meters; the entire circuit is 14 m, connected with a galvanized busbar buried ~60 cm deep, bolted joints, and a 6.5 mm² cable to a single grounded outlet used for my setup. It is located about 6 meters from my desk. I built a TT-like system after trying different TN-C and TN-C-S variations. In my case TT performed better subjectively, but this is only my observation. With TN-C-S, I had more desync, the same as with TN-C. However, using my own PE instead of PEN did not consistently change the problem. For the TN-C and TN-C-S tests, I disconnected the entire house wiring; I made a completely separate line for the PC outlet.

And none of this would have happened if it weren't for my 211 letters to all possible authorities in my country over the past two years, after discovering that the problem was directly related to the quality of the power supply to the entire line and the village as a whole.
And I had an radio-monitoring inspection at my place. And that's the response they sent me: "Based on the results of the measures carried out by employees of the Central Branch of the UCRF in your presence on the premises and in the rooms of the building, in the public radio frequency bands, including the frequency bands for cellular radio communications, terrestrial television broadcasting, terrestrial analog audio broadcasting, and broadband wireless radio access, there were no violations of the radio emission parameters of the specified radio technologies and unacceptable radio interference in the public radio frequency bands.
It was recorded that the PC system unit itself emits broadband radiation in the approximate range of 308-730 MHz with levels up to ~60 dBm.
In addition, we would like to inform you that the Central Branch of the Ukralnlan State Centre for Radio Frequencies has not received any requests from operators providing telecommunications services in the village of Yours or from other residents of the specified area to identify and eliminate sources of radio interference that may cause reception problems in public radio frequency bands."
I am attaching a file of the device used to take measurements at my place via the Central Branch of the UCRF.
I also invited a metrologist who brought a C1-67 universal oscilloscope, and we checked the sinusoid, and everything was fine with it.

My current hypothesis is that several mechanisms may coexist: unstable PEN, voltage unbalance, common-mode currents, differential-mode conducted EMI, neutral impedance, and grounding reference instability.

I also created a bunch of threads on other forums. Here's one of the last ones, where there was some discussion but no results, so if you're bored, you can read it.

https://www.eevblog.com/forum/beginners ... l#lastPost

Network and routing layer observations

I also need to include the network layer, because it cannot be ignored in this system.
I have two fiber providers at home. Both are pure fiber connections with no copper involved.
Depending on the provider, the in-game experience changes significantly. This includes perceived responsiveness, animation timing, enemy appearance timing, and general synchronization of gameplay events.
Enemy peeks, movement speed perception, hit registration consistency, and overall “game pacing” feel different depending on the ISP.
This is not a subtle difference. It feels like a different network simulation of the same game.
I also tested different routers, including OpenWRT configurations. I experimented with queue management (SQM), bufferbloat tuning, and different traffic shaping approaches. Results were mixed: sometimes improvements, sometimes degradation. I would not consider it a stable fix, but it can influence perceived consistency in some cases.
Static IP did not resolve the issue.
In 2024 (August-October), I also extensively tested VPS-based routing setups. I purchased and tested a large number of VPS instances across Germany, Poland, and Ukralne.
Most tests were done using WireGuard tunnels over Linux. This introduced additional variability, since performance differed heavily between servers.
Later I also experimented with alternatives such as FastD and GRE tunneling. GRE showed better stability in theory..
Across different servers and routes, the in-game experience changed drastically. Each endpoint produced a noticeably different “game feel”, including movement perception, timing, and hit registration consistency.
At this point, I cannot fully exclude the network layer as a contributing factor, especially in online environments.
However, I also cannot isolate it as the sole cause, since offline behavior is still affected.

I am also an admin in the Discord group “noinput group" under the nickname fa1kk. I’m sure most of the people here know about that discord, so I don’t think the link is required. If someone wants to reach out or discuss findings, please do so in public channels only via this forum topic or discord; I’m not answering DMs.

These videos are not meant to prove a cause, only to document repeatable behavior differences across sessions( I have many more, but I think those are more than enough).

https://youtu.be/XYyGBdzw20s - Clips #1 - Valorant DM and Swiftplay (Reminder how cool it is to play with this kind of issues).

https://www.youtube.com/watch?v=x4E2jcEzSOc - Clips #2 – Valorant DM Transition / switching behavior from ok state to bad state, heavy aim feel, increased blur perception.

https://www.youtube.com/watch?v=j49P1EhqLzM - Clips #3 – Valorant DM just a bad state.

Here I attached
Spectrum analyzer that Used by UCRF inspection;
Power quality measurements from DTEK (flicker, voltage unbalance, U0/U2 deviations) Phase imbalance and flicker graphs collected during monitoring period.These are included for reference, not as conclusions.

[cerns]

As far as I can tell, there isn't much left that you haven't tried. Is the situation the same when you test your PC at your friend's place? At this point, I'm actually more curious about the cause than about fixing the problem, lol.

[fa1kk]

As far as I can tell, there isn't much left that you haven't tried. Is the situation the same when you test your PC at your friend's place? At this point, I'm actually more curious about the cause than about fixing the problem, lol.

The PC itself has no issues, since the PC, including the monitor, is working fine in different places. Unfortunately, I don't have an option to move any more. So I'm stuck in my place for some time.

I even had a chance to be powered from another source rather than my Thermal power station, and the difference was there, but not that big of a deal.

One interesting observation came from testing the same PC while being supplied through three different 10 kV -0.4 kV distribution transformers.

The gameplay characteristics were not identical between them. However, the differences appeared to correlate much more with the instantaneous load on the feeder than with the transformer itself.

In other words, changing the transformer changed the behavior somewhat, but it did not eliminate the issue. This suggests that the transformer may not be the root cause.

Here I attached Transformer (THD)

[Asesino34]

Interesting post. It would be nice to know the characteristics of each ISP you've had so we could check whether an improvement is possible depending on certain factors.
For example, how much MTU fragmentation each ISP causes and how powerful the router they provide is. A couple of years ago, my connection worked well when I switched to another company. That ISP wasn't widely used where I live, and they gave me a router that honestly had some interesting specifications. We're talking about 512 MB of RAM, a dual-core CPU, and a different packet identification technology
My router eventually burned out, and they replaced it with another one, but the newer models were worse—half the RAM and lower overall performance. I can't say for sure that my problems started from that point onward, but what you're saying is definitely interesting

[fa1kk]

Interesting post. It would be nice to know the characteristics of each ISP you've had so we could check whether an improvement is possible depending on certain factors.
For example, how much MTU fragmentation each ISP causes and how powerful the router they provide is. A couple of years ago, my connection worked well when I switched to another company. That ISP wasn't widely used where I live, and they gave me a router that honestly had some interesting specifications. We're talking about 512 MB of RAM, a dual-core CPU, and a different packet identification technology
My router eventually burned out, and they replaced it with another one, but the newer models were worse—half the RAM and lower overall performance. I can't say for sure that my problems started from that point onward, but what you're saying is definitely interesting

In my case, I own all of the networking equipment, so I'm not limited by the ISP hardware.

The fiber goes directly into my own ONU, then either:
1)ONU -> 25 cm Cat5e -> PC, or
2) ONU -> 25 cm Cat5e -> my router -> 25 сm Cat5e -> PC.
I've tested both configurations. The router itself doesn't improve anything; The gameplay characteristics remain essentially the same.

Regarding MTU, all of the ISPs I've tested behave identically. The largest unfragmented ICMP packet is 1472 bytes (ping -f -l 1472 succeeds, 1473 fragments), so there is no meaningful difference there either of them.

The two ISPs I currently have are: AS21497 (UMC-AS) and AS57197 (SOYUZTELECOM-AS)

Gameplay is noticeably different between them, but I believe that's mainly due to differences in routing and peering.
Realistically, I don't think that an ISP would build a custom route for a single customer, since route changes affect traffic engineering for many users. That's why I don't consider routing a practical long-term solution, even though it clearly influences online gameplay.

[Asesino34]

Interesting post. It would be nice to know the characteristics of each ISP you've had so we could check whether an improvement is possible depending on certain factors.
For example, how much MTU fragmentation each ISP causes and how powerful the router they provide is. A couple of years ago, my connection worked well when I switched to another company. That ISP wasn't widely used where I live, and they gave me a router that honestly had some interesting specifications. We're talking about 512 MB of RAM, a dual-core CPU, and a different packet identification technology
My router eventually burned out, and they replaced it with another one, but the newer models were worse—half the RAM and lower overall performance. I can't say for sure that my problems started from that point onward, but what you're saying is definitely interesting

In my case, I own all of the networking equipment, so I'm not limited by the ISP hardware.

The fiber goes directly into my own ONU, then either:
1)ONU -> 25 cm Cat5e -> PC, or
2) ONU -> 25 cm Cat5e -> my router -> 25 сm Cat5e -> PC.
I've tested both configurations. The router itself doesn't improve anything; The gameplay characteristics remain essentially the same.

Regarding MTU, all of the ISPs I've tested behave identically. The largest unfragmented ICMP packet is 1472 bytes (ping -f -l 1472 succeeds, 1473 fragments), so there is no meaningful difference there either of them.

The two ISPs I currently have are: AS21497 (UMC-AS) and AS57197 (SOYUZTELECOM-AS)

Gameplay is noticeably different between them, but I believe that's mainly due to differences in routing and peering.
Realistically, I don't think that an ISP would build a custom route for a single customer, since route changes affect traffic engineering for many users. That's why I don't consider routing a practical long-term solution, even though it clearly influences online gameplay.

How well-known or widely used are those ISPs? Could user congestion at certain times of the day be taken into account?
Usually, people tend to choose one of the three most popular ISPs in their country. I'm not sure if anyone on this forum has tried a very small or relatively unknown ISP with maybe only around 100 customers
It's true that the routing may be the same, but the level of congestion on the ISP's systems and network infrastructure may not be

[fa1kk]

Interesting post. It would be nice to know the characteristics of each ISP you've had so we could check whether an improvement is possible depending on certain factors.
For example, how much MTU fragmentation each ISP causes and how powerful the router they provide is. A couple of years ago, my connection worked well when I switched to another company. That ISP wasn't widely used where I live, and they gave me a router that honestly had some interesting specifications. We're talking about 512 MB of RAM, a dual-core CPU, and a different packet identification technology
My router eventually burned out, and they replaced it with another one, but the newer models were worse—half the RAM and lower overall performance. I can't say for sure that my problems started from that point onward, but what you're saying is definitely interesting

In my case, I own all of the networking equipment, so I'm not limited by the ISP hardware.

The fiber goes directly into my own ONU, then either:
1)ONU -> 25 cm Cat5e -> PC, or
2) ONU -> 25 cm Cat5e -> my router -> 25 сm Cat5e -> PC.
I've tested both configurations. The router itself doesn't improve anything; The gameplay characteristics remain essentially the same.

Regarding MTU, all of the ISPs I've tested behave identically. The largest unfragmented ICMP packet is 1472 bytes (ping -f -l 1472 succeeds, 1473 fragments), so there is no meaningful difference there either of them.

The two ISPs I currently have are: AS21497 (UMC-AS) and AS57197 (SOYUZTELECOM-AS)

Gameplay is noticeably different between them, but I believe that's mainly due to differences in routing and peering.
Realistically, I don't think that an ISP would build a custom route for a single customer, since route changes affect traffic engineering for many users. That's why I don't consider routing a practical long-term solution, even though it clearly influences online gameplay.

How well-known or widely used are those ISPs? Could user congestion at certain times of the day be taken into account?
Usually, people tend to choose one of the three most popular ISPs in their country. I'm not sure if anyone on this forum has tried a very small or relatively unknown ISP with maybe only around 100 customers
It's true that the routing may be the same, but the level of congestion on the ISP's systems and network infrastructure may not be

I've actually tested both large national ISPs and much smaller local ones. One of the smaller providers had only around a thousand users, and according to their monitoring, the access network utilization was below roughly 12%. The gameplay didn't change in any way.

I've also played over Starlink and LTe, 5g over mobile data. The experience definitely changed depending on the connection quality. The most stable connection almost always felt the best.

What I've consistently noticed is that players with unstable connections appear unstable regardless of the game. I have a friend whose connection behaves like that. 9 out of 10 people who played against him are confirming that he often appears to teleport, finish peeking before the animation catches up, or kill before he visually completes the movement and barely starts peeking. At one point his ISP changed the routing, and afterwards he appeared completely different in-game without changing anything on his PC.

That makes me think routing and peering can significantly affect how online games feel, although I can't prove that they're more important than congestion in general.

Another interesting observation is that one of my ISPs completely replaced their access equipment after acquiring another company. Before that upgrade, purchasing a static public IP made the game super bad compared to using a dynamic private address. After the network modernization, both behave in the same way, which surprised me.

At the same time, the offline symptoms remain, which is why I don't think the network layer alone explains everything I'm seeing and only proves the point that the issue we are dealing with has layers to it, not just 1 specific source of the issue.

[Asesino34]

Interesting post. It would be nice to know the characteristics of each ISP you've had so we could check whether an improvement is possible depending on certain factors.
For example, how much MTU fragmentation each ISP causes and how powerful the router they provide is. A couple of years ago, my connection worked well when I switched to another company. That ISP wasn't widely used where I live, and they gave me a router that honestly had some interesting specifications. We're talking about 512 MB of RAM, a dual-core CPU, and a different packet identification technology
My router eventually burned out, and they replaced it with another one, but the newer models were worse—half the RAM and lower overall performance. I can't say for sure that my problems started from that point onward, but what you're saying is definitely interesting

In my case, I own all of the networking equipment, so I'm not limited by the ISP hardware.

The fiber goes directly into my own ONU, then either:
1)ONU -> 25 cm Cat5e -> PC, or
2) ONU -> 25 cm Cat5e -> my router -> 25 сm Cat5e -> PC.
I've tested both configurations. The router itself doesn't improve anything; The gameplay characteristics remain essentially the same.

Regarding MTU, all of the ISPs I've tested behave identically. The largest unfragmented ICMP packet is 1472 bytes (ping -f -l 1472 succeeds, 1473 fragments), so there is no meaningful difference there either of them.

The two ISPs I currently have are: AS21497 (UMC-AS) and AS57197 (SOYUZTELECOM-AS)

Gameplay is noticeably different between them, but I believe that's mainly due to differences in routing and peering.
Realistically, I don't think that an ISP would build a custom route for a single customer, since route changes affect traffic engineering for many users. That's why I don't consider routing a practical long-term solution, even though it clearly influences online gameplay.

How well-known or widely used are those ISPs? Could user congestion at certain times of the day be taken into account?
Usually, people tend to choose one of the three most popular ISPs in their country. I'm not sure if anyone on this forum has tried a very small or relatively unknown ISP with maybe only around 100 customers
It's true that the routing may be the same, but the level of congestion on the ISP's systems and network infrastructure may not be

I've actually tested both large national ISPs and much smaller local ones. One of the smaller providers had only around a thousand users, and according to their monitoring, the access network utilization was below roughly 12%. The gameplay didn't change in any way.

I've also played over Starlink and LTe, 5g over mobile data. The experience definitely changed depending on the connection quality. The most stable connection almost always felt the best.

What I've consistently noticed is that players with unstable connections appear unstable regardless of the game. I have a friend whose connection behaves like that. 9 out of 10 people who played against him are confirming that he often appears to teleport, finish peeking before the animation catches up, or kill before he visually completes the movement and barely starts peeking. At one point his ISP changed the routing, and afterwards he appeared completely different in-game without changing anything on his PC.

That makes me think routing and peering can significantly affect how online games feel, although I can't prove that they're more important than congestion in general.

Another interesting observation is that one of my ISPs completely replaced their access equipment after acquiring another company. Before that upgrade, purchasing a static public IP made the game super bad compared to using a dynamic private address. After the network modernization, both behave in the same way, which surprised me.

At the same time, the offline symptoms remain, which is why I don't think the network layer alone explains everything I'm seeing and only proves the point that the issue we are dealing with has layers to it, not just 1 specific source of the issue.

I can confirm the unstable connection part, although it's also possible that some of them simply have very little bandwidth available
I also wonder whether ISPs change anything depending on the speed tier they offer. These days, 300 Mbps or higher is pretty standard
There is clearly some kind of issue if most of us here either have connection problems or are at least obsessed with investigating the same thing.

Thanks for the reply

[fa1kk]

In my case, I own all of the networking equipment, so I'm not limited by the ISP hardware.

The fiber goes directly into my own ONU, then either:
1)ONU -> 25 cm Cat5e -> PC, or
2) ONU -> 25 cm Cat5e -> my router -> 25 сm Cat5e -> PC.
I've tested both configurations. The router itself doesn't improve anything; The gameplay characteristics remain essentially the same.

Regarding MTU, all of the ISPs I've tested behave identically. The largest unfragmented ICMP packet is 1472 bytes (ping -f -l 1472 succeeds, 1473 fragments), so there is no meaningful difference there either of them.

The two ISPs I currently have are: AS21497 (UMC-AS) and AS57197 (SOYUZTELECOM-AS)

Gameplay is noticeably different between them, but I believe that's mainly due to differences in routing and peering.
Realistically, I don't think that an ISP would build a custom route for a single customer, since route changes affect traffic engineering for many users. That's why I don't consider routing a practical long-term solution, even though it clearly influences online gameplay.

How well-known or widely used are those ISPs? Could user congestion at certain times of the day be taken into account?
Usually, people tend to choose one of the three most popular ISPs in their country. I'm not sure if anyone on this forum has tried a very small or relatively unknown ISP with maybe only around 100 customers
It's true that the routing may be the same, but the level of congestion on the ISP's systems and network infrastructure may not be

I've actually tested both large national ISPs and much smaller local ones. One of the smaller providers had only around a thousand users, and according to their monitoring, the access network utilization was below roughly 12%. The gameplay didn't change in any way.

I've also played over Starlink and LTe, 5g over mobile data. The experience definitely changed depending on the connection quality. The most stable connection almost always felt the best.

What I've consistently noticed is that players with unstable connections appear unstable regardless of the game. I have a friend whose connection behaves like that. 9 out of 10 people who played against him are confirming that he often appears to teleport, finish peeking before the animation catches up, or kill before he visually completes the movement and barely starts peeking. At one point his ISP changed the routing, and afterwards he appeared completely different in-game without changing anything on his PC.

That makes me think routing and peering can significantly affect how online games feel, although I can't prove that they're more important than congestion in general.

Another interesting observation is that one of my ISPs completely replaced their access equipment after acquiring another company. Before that upgrade, purchasing a static public IP made the game super bad compared to using a dynamic private address. After the network modernization, both behave in the same way, which surprised me.

At the same time, the offline symptoms remain, which is why I don't think the network layer alone explains everything I'm seeing and only proves the point that the issue we are dealing with has layers to it, not just 1 specific source of the issue.

I can confirm the unstable connection part, although it's also possible that some of them simply have very little bandwidth available
I also wonder whether ISPs change anything depending on the speed tier they offer. These days, 300 Mbps or higher is pretty standard
There is clearly some kind of issue if most of us here either have connection problems or are at least obsessed with investigating the same thing.

Thanks for the reply

I've tested different ISPs, different bandwidth tiers, and even different access technologies. In my case, simply having more or less bandwidth never changed anything. Almost any games use only a tiny fraction of that connection, so I don't think bandwidth itself is the limiting factor here.

Packet loss, jitter, and latency certainly affect the online experience, but they don't explain my entire situation. My packet loss and jitter are stable, yet the problem is still there. I need to emphasize that when there is no electricity in the city, both my ISPs are using different autonomous sources of power to keep the internet going. In any online game, I experience more desync than usual.

At this point, I think it's more likely that the ISP's infrastructure, routing, hardware, power conditions, and network configuration play a much larger role than the typical connection speed.

One interesting observation happened back in 2021. I had direct contact with a network engineer from my ISP. While I was playing, he was making routing or network adjustments on their side. The difference in-game was dramatic and stayed for quite a while. And after 2022 the ISP no longer allows customers to communicate with engineers beyond first-line support, so I couldn't continue those tests.

Another observation was that for about a day after purchasing the VPS servers. On every server, every match, and every game felt exceptionally good. Whatever was changed eventually disappeared after 1 day (thanks to BGP). Funny enough, I was using this server not alone but with friends, and they experienced the same amazing feeling of the game on any server, and suddenly, the next day, all of us couldn't even play using that server. And worth mentioning, all of those guys I'm talking about are living in different countries.

The difficulty is that Internet routing is constantly changing. Packets traverse multiple autonomous systems, data centers, and BGP decisions before reaching the game server. As a gamer, there is simply no practical way to observe or control every routing decision along the path.

That's why I think we need objective metrics outside of the games themselves. In-game statistics don't tell us much beyond ping and packet loss, while the actual gameplay experience can still vary dramatically.

Even the ISP engineers I've spoken lately to didn't know where to begin looking for this kind of problem.

Personally, I would focus on solving the offline symptoms first: mouse feel, display clarity, and general responsiveness. If those can be stabilized, then any remaining online issues can be investigated separately. Otherwise, we're trying to solve two very large problems at the same time.

[danbmx]

Great post! You've literally tried everything. Honestly, there might not be a solution to this issue—it's highly likely just dirty power coming from the grid. By the way, have you tried making the neutral-to-ground bond impedance as low as possible?