Stop patching a dying engine – The entire game needs a next-gen upgrade. It’s 2026.

Darque.Fluxb14a_ESO · June 17

Just throwing in a few cents of thought.

They have been resolving some technical debt recently, it shows in the various patch notes where they say things along the lines of "We changed some things in the background that should make things go faster and have no noticable effect". We've been getting a lot more of those.

Upgrading the client engine to a new one, is a ridiculously large task with minimal payout, I don't expect this to ever get done for ESO. If they do anything for it, I expect them to iterate on the current engine. This approach is also less likely to cause massive amounts of regression bugs.

You have been saying "New-Gen" but that term is typically used to refer to newer gaming consoles. If we are talking about server upgrades to "New-Gen" it would be the evolution of matrix multiplication accumulation computational hardware (The stuff used in ML inference and training), and Upgrading the servers to use this would be a very technical undertaking that could provide significant improve performance, but would risk incredible amounts of regression bugs all over that could be very hard to detect, reproduce, and fix.

TLDR; I hope they can put more of their budget into the technical debt resolution that they seem to have been doing. Upgrading to a new engine is generally a dumb idea for a mature product.

VidmaVirtual · June 17

To make it simpler for those who think this is just an unrealistic demand for "prettier graphics," let’s look at the core issue through a simple everyday metaphor:

Imagine we own a 15-year-old house. The plumbing and the electrical wiring hidden inside the walls are ancient and weak. Over the years, we keep buying brand-new, expensive appliances—a top-of-the-line refrigerator, a powerful dishwasher, and a modern microwave (which are the game's new expansions, chapters, and mechanics). But the moment we try to turn them all on at the same time during peak hours, the house blows a fuse and the power goes out (the server chokes, causing heavy prime-time lag and massive queues for Gray Host).

No matter how many expensive appliances we buy for that kitchen, nothing will ever run smoothly until the owners completely open up the walls and upgrade the underlying wires and pipes.

This isn't about making the game look prettier; it’s about fixing the "wiring" so the game can actually support modern player populations, crossplay, and complex combat calculations simultaneously. Opening empty overflow campaigns like Vengeance is just like turning off half the appliances so the fuse doesn't blow—it doesn't fix the house, it just limits how we live in it.

We have supported this house for over a decade with our subscriptions and purchases. It is completely reasonable to ask the owners (ZOS/Microsoft) to invest in a solid, modern foundation so this game can live and thrive for another 10 years or more.

Darque.Fluxb14a_ESO · June 17

VidmaVirtual wrote: »

To make it simpler for those who think this is just an unrealistic demand for "prettier graphics," let’s look at the core issue through a simple everyday metaphor:

Imagine we own a 15-year-old house.

Your metaphor is a massive understatement of how much is going on, a closer metaphor would be that you have a 100-year-old city, and your recommendation is to tear the whole thing down, to rebuild the electical, sewage, and water systems. It could be done, it would cost a massive amount of money and time, the risk that you would *** up some of the existing behaviors is nearly inevitable as well.

With a 100-year-old city, you carefully approach the rebuilding of each system section by section, step by step.

VidmaVirtual · June 17

@Darque.Fluxb14a_ESO

I actually love your 100-year-old city metaphor, and I agree it captures the sheer scale of the problem much better than my kitchen example!

But if we look closely at how modern urban planning deals with a 100-year-old city, there comes a critical tipping point. When the population grows and technology changes, you cannot just keep fixing individual pipes section by section forever. If the main sewage line under the entire city center is completely collapsed, or if the central power grid cannot handle modern electrical demands, patching one street at a time won't prevent the citywide blackouts.

At some point, city leadership has to make the heavy, expensive decision to dig up the central avenues and completely replace the core infrastructure grid to sustain the next century of growth. Yes, it causes major disruption, it costs a massive amount of money, and it carries huge engineering risks. But the alternative is watching the city slowly suffocate under its own weight and decay.

That is exactly where ESO is in 2026. The "central grid" of the engine—the single-threaded calculation bottleneck—is currently overloaded. ZOS has been doing exactly what you suggested: approaching it step-by-step and section-by-section for years. But despite all those section patches, we are still stuck with severely reduced campaign populations and frustrating prime-time queues just to keep the "city" from collapsing.

Incremental maintenance is great for extending a product's life, but when you are looking at the next 10 years of survival, sometimes digging up the main roads and replacing the core foundation becomes a necessity, not just a preference.

SneaK · June 17

Darque.Fluxb14a_ESO wrote: »

VidmaVirtual wrote: »

To make it simpler for those who think this is just an unrealistic demand for "prettier graphics," let’s look at the core issue through a simple everyday metaphor:

Imagine we own a 15-year-old house.

Your metaphor is a massive understatement of how much is going on, a closer metaphor would be that you have a 100-year-old city, and your recommendation is to tear the whole thing down, to rebuild the electical, sewage, and water systems. It could be done, it would cost a massive amount of money and time, the risk that you would *** up some of the existing behaviors is nearly inevitable as well.

With a 100-year-old city, you carefully approach the rebuilding of each system section by section, step by step.

Right….. but they still upgrade the piping and bring the electrical up to code. And the new additions aren’t using the same 100 year materials either. Not really a “closer metaphor” at all.

sarahthes · June 17

The other game I play did completely relaunch their game after upgrading to a brand new engine.

It has taken them over 3 years already and they have only converted and re-released about 70% of their previous content, and they've had to aggressively monetize and add microtransactions to pay for the ongoing development. They lost a significant portion of their paying customer base, as well - the legacy game which is no longer supported still maintains a population roughly equal to the pop on the updated engine. So the new version has lower pop in general than the old game did at its prime.

I don't think this is something ESO should do.

VidmaVirtual · June 17

sarahthes wrote: »

The other game I play did completely relaunch their game after upgrading to a brand new engine.

It has taken them over 3 years already and they have only converted and re-released about 70% of their previous content, and they've had to aggressively monetize and add microtransactions to pay for the ongoing development. They lost a significant portion of their paying customer base, as well - the legacy game which is no longer supported still maintains a population roughly equal to the pop on the updated engine. So the new version has lower pop in general than the old game did at its prime.

I don't think this is something ESO should do.

Hi sarahthes!
Out of curiosity, which game was that? I’d love to read more about how their engine migration went and what specific challenges they faced.

kojou · June 17

I can see where you and your LLM are coming from, but I don't see Zenimax making either the monetary investment or the time investment required.

If they could somehow sell it as ESO 2.0 and have the community pay for the upgrade then maybe that could work, but would everyone pay the equivalent to at least a annual chapter to have the same game on the new engine? Would they want to play through all the new bugs that are likely to be introduced? The advantages would have to be very compelling, and I don't think most players would find it worth it. Maybe they could issue another survey to confirm...

I would also expect that the developers have access to LLMs to look at the code and advise as well, so if the only way out of a mess of code was an engine redesign/reimplementation their AI would tell them the same thing your LLM does and they could do a cost/benefit analysis...

VidmaVirtual · June 17

kojou wrote: »

I can see where you and your LLM are coming from, but I don't see Zenimax making either the monetary investment or the time investment required.

If they could somehow sell it as ESO 2.0 and have the community pay for the upgrade then maybe that could work, but would everyone pay the equivalent to at least a annual chapter to have the same game on the new engine? Would they want to play through all the new bugs that are likely to be introduced? The advantages would have to be very compelling, and I don't think most players would find it worth it. Maybe they could issue another survey to confirm...

I would also expect that the developers have access to LLMs to look at the code and advise as well, so if the only way out of a mess of code was an engine redesign/reimplementation their AI would tell them the same thing your LLM does and they could do a cost/benefit analysis...

@kojou Thank you for a very balanced and fair response! I actually agree with you that ZOS developers probably use LLMs for code analysis and their internal metrics likely tell them the exact same thing.

Regarding the "ESO 2.0" dilemma and whether players would pay for it: I don't think players should have to pay a single cent extra for a core infrastructure upgrade. A technical overhaul shouldn't be packaged as a new commercial product or a paid DLC. It should be funded directly from the massive ongoing revenue that the community already provides through subscriptions, chapters, and Crown Store spending.

As mentioned earlier in this thread, ESO brings in an average of $15 million every single month. In the software and live-service industry, reinvesting a portion of those profits back into the core architecture isn't charity—it's necessary product asset management. If a multi-billion-dollar car company realizes their main factory assembly line is outdated and limiting production, they don't ask their current car owners to pay for a new factory; they fund it from their corporate capital to stay competitive.

You are completely right that a new engine would bring new bugs initially. That is a guaranteed risk. But the compelling advantage that would make it worth it for the average player isn't "prettier graphics"—it's the elimination of 1-2 hour prime-time queues, fluid combat responsiveness where skills actually register during mass sieges, and a stable foundation that can seamlessly handle upcoming features like crossplay.

A survey would actually be a fascinating idea! I think if ZOS asked the community: "Would you prefer a smaller yearly chapter if it meant the team spent those resources on a massive performance and engine overhaul?", the response from veteran players might surprise them.

Gabriel_H · June 17

VidmaVirtual wrote: »

Gabriel_H wrote: »

VidmaVirtual wrote: »

@Gabriel_H

But what actually executes the ruleset? The engine code.

No. Again, that isn't the engine. Lets stick with the car metaphor.

The engine makes the whole thing go. The ruleset is the steering, brakes, lights and indicators.

ZOS have optimised the game code (the ruleset) as much as they can. Without inventing an entirely new programming language that is more efficient than C++ they've taken that as far as it can go.

The servers cannot process efficiently all the calculations the game code is throwing at them. There is no getting around that without improving the speeds of server processors, both single and multi-threaded - which isn't up to ZOS.

So, ZOS are left with two choices: Reduce the population further or revise the ruleset. They went with the latter and offered up Vengeance.

I don't like Vengeance, not least because I believe there was a more elegant solution. Me being picky aside, I feel it is too limiting and can be expanded without affecting performance - because the performance is better, a lot lot better.

So, which do you want? Less people or less calculations?

Gabriel_H
Your car metaphor actually highlights exactly where the logic fails.

The "ruleset" (steering, brakes, lights) just tells the car what to do. But the game engine is the transmission and the chassis that transfers that power to the wheels. If you put a modern V8 server processor into a car with a 1990s transmission, you cannot shift gears efficiently. No matter how much power the server has, the outdated engine architecture limits how that power is used.

Also, saying we need a new programming language better than C++ is completely incorrect. The world's most powerful, high-performance game engines—including Unreal Engine 5 and the backends of modern, massive multiplayer architectures—are written entirely in C++. The problem isn't the C++ language itself; it is the legacy, single-threaded architecture of *this specific engine's code* written over a decade ago.

Modern C++ development utilizes highly advanced parallel programming paradigms, asynchronous data pipelines, and Entity Component Systems (ECS) that can natively map calculations across dozens of modern CPU cores simultaneously without data corruption.

So to answer your question: we do not have to choose between "less people or less calculations." That is a false dilemma based on 2010-era software limits. With a modern next-gen engine architecture written in modern C++, the server can easily handle *more people* doing *more calculations* simultaneously. The limitation is the software framework, not the laws of physics or the programming language.

No. The engine is the engine. Period.

The game code is the rest of the car.

The game code isn't single threaded. Player actions are because they have to be. Even multi-threaded calculations have to have a single threaded master thread. Multi-threading has to have locks to stop data corruption and race conditions.

You are imagining a fantasy scenario.

The bottleneck is and will always be player actions. Multi-threads do not solve that.

Gabriel_H · June 17

SneaK wrote: »

I cannot see how anyone would argue against OPs suggestion. Like why on earth would we not want ZOS to upgrade their tech?

Because it doesn't solve the problem. It's a waste of time, resources and money.

SneaK · June 17

Gabriel_H wrote: »

SneaK wrote: »

I cannot see how anyone would argue against OPs suggestion. Like why on earth would we not want ZOS to upgrade their tech?

Because it doesn't solve the problem. It's a waste of time, resources and money.

So you’re saying the human race in 2026 is not capable of having more than 360 players fighting each other with standard ruleset in Cyro? Interesting.

VidmaVirtual · June 17

Gabriel_H wrote: »

VidmaVirtual wrote: »

Gabriel_H wrote: »

VidmaVirtual wrote: »

@Gabriel_H

But what actually executes the ruleset? The engine code.

No. Again, that isn't the engine. Lets stick with the car metaphor.

The engine makes the whole thing go. The ruleset is the steering, brakes, lights and indicators.

ZOS have optimised the game code (the ruleset) as much as they can. Without inventing an entirely new programming language that is more efficient than C++ they've taken that as far as it can go.

The servers cannot process efficiently all the calculations the game code is throwing at them. There is no getting around that without improving the speeds of server processors, both single and multi-threaded - which isn't up to ZOS.

So, ZOS are left with two choices: Reduce the population further or revise the ruleset. They went with the latter and offered up Vengeance.

I don't like Vengeance, not least because I believe there was a more elegant solution. Me being picky aside, I feel it is too limiting and can be expanded without affecting performance - because the performance is better, a lot lot better.

So, which do you want? Less people or less calculations?

Gabriel_H
Your car metaphor actually highlights exactly where the logic fails.

The "ruleset" (steering, brakes, lights) just tells the car what to do. But the game engine is the transmission and the chassis that transfers that power to the wheels. If you put a modern V8 server processor into a car with a 1990s transmission, you cannot shift gears efficiently. No matter how much power the server has, the outdated engine architecture limits how that power is used.

Also, saying we need a new programming language better than C++ is completely incorrect. The world's most powerful, high-performance game engines—including Unreal Engine 5 and the backends of modern, massive multiplayer architectures—are written entirely in C++. The problem isn't the C++ language itself; it is the legacy, single-threaded architecture of *this specific engine's code* written over a decade ago.

Modern C++ development utilizes highly advanced parallel programming paradigms, asynchronous data pipelines, and Entity Component Systems (ECS) that can natively map calculations across dozens of modern CPU cores simultaneously without data corruption.

So to answer your question: we do not have to choose between "less people or less calculations." That is a false dilemma based on 2010-era software limits. With a modern next-gen engine architecture written in modern C++, the server can easily handle *more people* doing *more calculations* simultaneously. The limitation is the software framework, not the laws of physics or the programming language.

No. The engine is the engine. Period.

The game code is the rest of the car.

The game code isn't single threaded. Player actions are because they have to be. Even multi-threaded calculations have to have a single threaded master thread. Multi-threading has to have locks to stop data corruption and race conditions.

You are imagining a fantasy scenario.

The bottleneck is and will always be player actions. Multi-threads do not solve that.

@Gabriel_H
No one is arguing that sync points or a main thread orchestrator don't exist in multi-threading. Of course they do. But there is a massive architectural difference between a single master thread waiting for synchronous worker threads, and a modern data-oriented framework built for massive concurrency.

Your argument that player actions must be entirely sequential on a single thread to prevent race conditions is based on legacy, object-oriented MMO architecture. Modern next-gen MMO backends solve this challenge using Deterministic Lockstep, Entity Component Systems (ECS), and Spatial Partitioning.

In a modern ECS framework, player actions are not processed as one heavy monolithic block. The server breaks down the data: positions, cooldowns, area-of-effect zones, and stat modifications are isolated in flat memory tables. Because these data components are separated, different CPU cores can calculate thousands of independent player variables (like stamina regeneration ticks or physics trajectory) simultaneously during the same server tick. They do not cause race conditions because they do not access the same memory addresses.

Furthermore, with spatial partitioning and modern server meshing, the network doesn't force one master thread to calculate all of Cyrodiil. The map is dynamically split into grid zones managed by independent server nodes. If a battle happens at Chalman Keep, that specific simulation is isolated to a dedicated node, preventing it from lagging out the entire campaign.

This isn't a "fantasy scenario"—it is the exact architecture being deployed in modern, high-population multiplayer networks in 2026. The bottleneck isn't the laws of physics or "player actions"; the bottleneck is a 2012 software framework that cannot distribute those actions across modern multi-core processors.

wolfie1.0. · June 17

VidmaVirtual wrote: »

@Aislinna

That is not how business accounting works. You cannot amortize active live-service revenue over pre-launch development years when the product wasn't even on the market generating sales.

The initial development costs from 2007 to 2014 were capital investments that were cleared and paid off a long time ago within the game's first few successful years.

We are discussing the current financial health of the game in 2026. Evaluating a live-service title's ability to fund a modern upgrade by dragging in pre-launch sunk costs from nearly two decades ago is a massive reach. The reality remains unchanged: the game generates immense monthly revenue right now, and reinvesting in its core technology is standard product lifecycle management.

Yes, lets talk about the curent financial health of ZOS.

You have to realize some realities here. You picked the absolute worst time to purpose a full rebuild. ZOS does not operate in a bubble.

First of all from what i understand based on my limited knowledge of GAAP you can Amortize developement expenses over the course of the anticipated lifetime of software once its launched to the public. So the cost wouldn't start really decreasing until 2014. But the real question isn't how well ESO and ZOS HAVE performed, its how they are doing now. So past earnings and figures prior to 2021 honestly dont matter. Once the Microsoft aquisition occurred how well ESO and ZOS preformed prior to then only matter in the sense that Microsoft was sold on an expected rate of performance and Return on Investment.

So lets talk corporate. $7.5 billion USD, that is the important number. That is what Microsoft paid to Zenimax private owners to aquire Zenimax, its publishers and game studios. It finalized in 2021, and folded it under the Xbox segment of the corporation. There is no denying that Microsoft could afford it, and could afford to take a loss on it, but corporate and the shareholders would love to see that investment provide an solid return.

So, lets look at the state of Xbox... its a mess, i wont bring up the details here but the news is everywhere. Xbox is NOT doing well. Projects, Studios, and games are folding. Consoles are NOT selling. Gamepass has gone through several pricing changes. Leaderhip has changed. Microsoft has redirected to AI and has invested unreal amounts of money in it.

Meanwhile Xbox is under new leadership, and while not publicy stated the implication is pretty obvious, turn a profit or be gone. New leaders have stated that they are going to invest in core IPs and properties. Thankfully ESO is technically a core IP, but its not mainline Elderscrolls, that could be a boon, it could be a problem, or it could be status qou. Keep in mind that ZOS was forced to cancel a project for a new game that was estimated to have already invested around $400 million (if not more) and created a leadership change. ZOS still needs to recover those funds.

All of this is happening in an enviroment where silicon costs have risen to extreme levels due to AI data center demands. RAM, GPUs, and Storage costs have increased substantially. RAM and storage 2 years ago would have been the cheapest components of any build (including consoles). We have seen price increases on EVERY major system for tech that is 4-5 years old. If Stadia launched today, it might actually have had a better chance to live on longer than it did.

Its in THIS enviroment that you want them to update the core code. essentially rebuild the game from almost scratch, and then relaunch the game, to what would be a smaller audience. Its simply not going to happen, at least not yet.

Honestly, ESO will likely be held to together as best it can until revenue catches up with the changes and recent upheavals in the industry. The Focus for Xbox is to get both a new Elderscrolls game, and likely a new Fallout game launched first. Then IF those are successful enough and IF Xbox shows a profit, then and only then do I see a possibility of getting a new engine and full build of the game.

until then, i think we should be a tad grateful that we still have a game to play and devs that are actually trying to improve the system. Some other live service game communities are not so lucky.

VidmaVirtual · June 17

wolfie1.0. wrote: »

VidmaVirtual wrote: »

@Aislinna

That is not how business accounting works. You cannot amortize active live-service revenue over pre-launch development years when the product wasn't even on the market generating sales.

The initial development costs from 2007 to 2014 were capital investments that were cleared and paid off a long time ago within the game's first few successful years.

We are discussing the current financial health of the game in 2026. Evaluating a live-service title's ability to fund a modern upgrade by dragging in pre-launch sunk costs from nearly two decades ago is a massive reach. The reality remains unchanged: the game generates immense monthly revenue right now, and reinvesting in its core technology is standard product lifecycle management.

Yes, lets talk about the curent financial health of ZOS.

You have to realize some realities here. You picked the absolute worst time to purpose a full rebuild. ZOS does not operate in a bubble.

First of all from what i understand based on my limited knowledge of GAAP you can Amortize developement expenses over the course of the anticipated lifetime of software once its launched to the public. So the cost wouldn't start really decreasing until 2014. But the real question isn't how well ESO and ZOS HAVE performed, its how they are doing now. So past earnings and figures prior to 2021 honestly dont matter. Once the Microsoft aquisition occurred how well ESO and ZOS preformed prior to then only matter in the sense that Microsoft was sold on an expected rate of performance and Return on Investment.

So lets talk corporate. $7.5 billion USD, that is the important number. That is what Microsoft paid to Zenimax private owners to aquire Zenimax, its publishers and game studios. It finalized in 2021, and folded it under the Xbox segment of the corporation. There is no denying that Microsoft could afford it, and could afford to take a loss on it, but corporate and the shareholders would love to see that investment provide an solid return.

So, lets look at the state of Xbox... its a mess, i wont bring up the details here but the news is everywhere. Xbox is NOT doing well. Projects, Studios, and games are folding. Consoles are NOT selling. Gamepass has gone through several pricing changes. Leaderhip has changed. Microsoft has redirected to AI and has invested unreal amounts of money in it.

Meanwhile Xbox is under new leadership, and while not publicy stated the implication is pretty obvious, turn a profit or be gone. New leaders have stated that they are going to invest in core IPs and properties. Thankfully ESO is technically a core IP, but its not mainline Elderscrolls, that could be a boon, it could be a problem, or it could be status qou. Keep in mind that ZOS was forced to cancel a project for a new game that was estimated to have already invested around $400 million (if not more) and created a leadership change. ZOS still needs to recover those funds.

All of this is happening in an enviroment where silicon costs have risen to extreme levels due to AI data center demands. RAM, GPUs, and Storage costs have increased substantially. RAM and storage 2 years ago would have been the cheapest components of any build (including consoles). We have seen price increases on EVERY major system for tech that is 4-5 years old. If Stadia launched today, it might actually have had a better chance to live on longer than it did.

Its in THIS enviroment that you want them to update the core code. essentially rebuild the game from almost scratch, and then relaunch the game, to what would be a smaller audience. Its simply not going to happen, at least not yet.

Honestly, ESO will likely be held to together as best it can until revenue catches up with the changes and recent upheavals in the industry. The Focus for Xbox is to get both a new Elderscrolls game, and likely a new Fallout game launched first. Then IF those are successful enough and IF Xbox shows a profit, then and only then do I see a possibility of getting a new engine and full build of the game.

until then, i think we should be a tad grateful that we still have a game to play and devs that are actually trying to improve the system. Some other live service game communities are not so lucky.

@wolfie1.0.
Thank you for this incredibly comprehensive and highly accurate macroeconomic breakdown. You are entirely correct about the current corporate climate, and I really appreciate you bringing these hard realities into the discussion.

The Xbox segment is indeed facing massive pressure, the rising hardware/silicon costs due to the AI boom are real, and ZOS cancelling a $400+ million project undoubtedly created a massive financial wound that the studio is desperately trying to heal. You are 100% right that from a traditional corporate standpoint, picking this exact moment to pitch a high-risk, expensive engine rebuild looks impossible.

However, if we look at it from the perspective of Microsoft’s expected Return on Investment (ROI), there is a counter-argument to consider: the cost of letting a core premium asset slowly suffocate.

As you mentioned, ESO is a core IP property that brings in highly reliable monthly recurring revenue. But that revenue relies entirely on player retention. When long-time players, high-end trial groups, and massive PvP guilds gradually walk away because the technical stagnation makes prime-time gameplay unplayable, Microsoft's expected rate of performance begins to decay.

Neglecting the core infrastructure to save short-term capital is how massive live-service games enter a terminal downward spiral. If the engine is currently so bottlenecked that ZOS has to severely slash campaign population limits just to keep the servers from crashing, they are fighting an uphill battle against their own product's expiration date.

While I completely agree with you that Xbox leadership is being hyper-reactive and focusing heavily on the upcoming mainline Elder Scrolls and Fallout titles, allowing their most reliable, active MMO to slowly decline due to technical debt is a dangerous strategy.

You made a brilliant point that we should be grateful to still have a live game with active developers—and I truly am. I love this game, which is why I’ve supported it for 11 years. But it’s precisely because I want to see it survive another decade under Microsoft's messy corporate umbrella that I believe a long-term architectural solution needs to be on the table, even if the current industry climate makes it incredibly difficult. Thank you for such an amazing contribution to this thread!

Gabriel_H · June 17

VidmaVirtual wrote: »

Your argument that player actions must be entirely sequential on a single thread to prevent race conditions is based on legacy, object-oriented MMO architecture. Modern next-gen MMO backends solve this challenge using Deterministic Lockstep, Entity Component Systems (ECS), and Spatial Partitioning.

That isn't my argument. My argument is that player actions ARE sequential. That has to run through a single thread. I'm trying to keep this simple as you can't even grasp the difference between engine, game code and server.

So lets complicate it up.

VidmaVirtual wrote: »

Modern next-gen MMO backends solve this challenge using Deterministic Lockstep, Entity Component Systems (ECS), and Spatial Partitioning

Deterministic Lockstep: Is client A to client B, rather than Client A to Server to Client B. The latency of that is determined by the player with the HIGHEST latency. So someone in Australia with a poor ISP running at 500ms latency means EVERYONE has high latency, freezes and delays.

Entity Component Systems: This could boost performance, but the bottleneck remains the calculations. Separating the object to asset & data doesn't change the number of calculations on that data.

Spatial Partitioning: ESO already has this. That's what instances are, and what phasing is. Doing it on the fly in a map the size of Cyrodiil would require splitting the map into multiple grids. With so many dynamic rather than static components this would add not reduce the load on the servers.

Gabriel_H · June 17

SneaK wrote: »

Gabriel_H wrote: »

SneaK wrote: »

I cannot see how anyone would argue against OPs suggestion. Like why on earth would we not want ZOS to upgrade their tech?

Because it doesn't solve the problem. It's a waste of time, resources and money.

So you’re saying the human race in 2026 is not capable of having more than 360 players fighting each other with standard ruleset in Cyro? Interesting.

No, I'm saying the ruleset is too complicated. There are too many calculations. Technology is not infallible and it has limits. As does any company who are required to show a decent ROI to a parent company who would shut the doors on them.

VidmaVirtual · June 17

Gabriel_H wrote: »

VidmaVirtual wrote: »

Your argument that player actions must be entirely sequential on a single thread to prevent race conditions is based on legacy, object-oriented MMO architecture. Modern next-gen MMO backends solve this challenge using Deterministic Lockstep, Entity Component Systems (ECS), and Spatial Partitioning.

That isn't my argument. My argument is that player actions ARE sequential. That has to run through a single thread. I'm trying to keep this simple as you can't even grasp the difference between engine, game code and server.

So lets complicate it up.

VidmaVirtual wrote: »

Modern next-gen MMO backends solve this challenge using Deterministic Lockstep, Entity Component Systems (ECS), and Spatial Partitioning

Deterministic Lockstep: Is client A to client B, rather than Client A to Server to Client B. The latency of that is determined by the player with the HIGHEST latency. So someone in Australia with a poor ISP running at 500ms latency means EVERYONE has high latency, freezes and delays.

Entity Component Systems: This could boost performance, but the bottleneck remains the calculations. Separating the object to asset & data doesn't change the number of calculations on that data.

Spatial Partitioning: ESO already has this. That's what instances are, and what phasing is. Doing it on the fly in a map the size of Cyrodiil would require splitting the map into multiple grids. With so many dynamic rather than static components this would add not reduce the load on the servers.

@Gabriel_H
If you are going to condescendingly tell me I "cannot grasp" technical concepts, you should at least make sure your own definitions are accurate, because almost everything you just wrote is fundamentally wrong.

1. Deterministic Lockstep: You are completely confusing modern server-authoritative determinism with 1990s Peer-to-Peer (P2P) network code used in old RTS games like Starcraft 1. Modern server determinism does NOT slow down the entire simulation for a high-ping player in Australia; the server dictates the master state tick, processes the parallelized inputs, and rolls back or drops late packages without affecting other players.

2. ECS (Entity Component System): You claim separating data from logic doesn't change the number of calculations. True, the *number* of calculations stays the same, but *how* they are executed changes completely. Instead of a single master thread processing heavy player objects one-by-one in a long slow line (sequential), an ECS arranges data linearly in memory cache, allowing modern multi-core server CPUs to process those calculations simultaneously (parallel computing). 32 cores doing 1000 calculations at the same time is lightyears faster than 1 core doing 32,000 calculations sequentially. That is how bottlenecks are broken.

3. Spatial Partitioning: Phasing and instancing are NOT spatial partitioning. Phasing puts players in entirely different visual realities. True Dynamic Spatial Partitioning (like server meshing used in modern network infrastructure) dynamically chops a live zone into virtual grid cells managed by different hardware nodes on the fly. It doesn't add load—it drastically reduces it by isolating the combat calculations of a castle siege so it doesn't lag out the rest of the map.

But as I promised myself before writing this: this will be my final response in this thread.

Thank you so much to everyone who jumped into this discussion over the last few days—including you, Gabriel_H, for keeping the debate lively. I really appreciate all the deep technical breakdowns, the macroeconomic insights, and the different perspectives we shared.

I have fully laid out my views as an 11-year veteran of ESO. Since these forum debates can easily spin in circles forever, I am stepping back to actually enjoy playing the game instead of analyzing its code. I'll leave the thread open for the community to continue the conversation.

See you all in Tamriel, and have a great time gaming! 🫶

Gabriel_H · June 17

VidmaVirtual wrote: »

Gabriel_H wrote: »

VidmaVirtual wrote: »

Your argument that player actions must be entirely sequential on a single thread to prevent race conditions is based on legacy, object-oriented MMO architecture. Modern next-gen MMO backends solve this challenge using Deterministic Lockstep, Entity Component Systems (ECS), and Spatial Partitioning.

That isn't my argument. My argument is that player actions ARE sequential. That has to run through a single thread. I'm trying to keep this simple as you can't even grasp the difference between engine, game code and server.

So lets complicate it up.

VidmaVirtual wrote: »

Modern next-gen MMO backends solve this challenge using Deterministic Lockstep, Entity Component Systems (ECS), and Spatial Partitioning

Deterministic Lockstep: Is client A to client B, rather than Client A to Server to Client B. The latency of that is determined by the player with the HIGHEST latency. So someone in Australia with a poor ISP running at 500ms latency means EVERYONE has high latency, freezes and delays.

Entity Component Systems: This could boost performance, but the bottleneck remains the calculations. Separating the object to asset & data doesn't change the number of calculations on that data.

Spatial Partitioning: ESO already has this. That's what instances are, and what phasing is. Doing it on the fly in a map the size of Cyrodiil would require splitting the map into multiple grids. With so many dynamic rather than static components this would add not reduce the load on the servers.

@Gabriel_H
If you are going to condescendingly tell me I "cannot grasp" technical concepts, you should at least make sure your own definitions are accurate, because almost everything you just wrote is fundamentally wrong.

1. Deterministic Lockstep: You are completely confusing modern server-authoritative determinism with 1990s Peer-to-Peer (P2P) network code used in old RTS games like Starcraft 1. Modern server determinism does NOT slow down the entire simulation for a high-ping player in Australia; the server dictates the master state tick, processes the parallelized inputs, and rolls back or drops late packages without affecting other players.

2. ECS (Entity Component System): You claim separating data from logic doesn't change the number of calculations. True, the *number* of calculations stays the same, but *how* they are executed changes completely. Instead of a single master thread processing heavy player objects one-by-one in a long slow line (sequential), an ECS arranges data linearly in memory cache, allowing modern multi-core server CPUs to process those calculations simultaneously (parallel computing). 32 cores doing 1000 calculations at the same time is lightyears faster than 1 core doing 32,000 calculations sequentially. That is how bottlenecks are broken.

3. Spatial Partitioning: Phasing and instancing are NOT spatial partitioning. Phasing puts players in entirely different visual realities. True Dynamic Spatial Partitioning (like server meshing used in modern network infrastructure) dynamically chops a live zone into virtual grid cells managed by different hardware nodes on the fly. It doesn't add load—it drastically reduces it by isolating the combat calculations of a castle siege so it doesn't lag out the rest of the map.

But as I promised myself before writing this: this will be my final response in this thread.

Thank you so much to everyone who jumped into this discussion over the last few days—including you, Gabriel_H, for keeping the debate lively. I really appreciate all the deep technical breakdowns, the macroeconomic insights, and the different perspectives we shared.

I have fully laid out my views as an 11-year veteran of ESO. Since these forum debates can easily spin in circles forever, I am stepping back to actually enjoy playing the game instead of analyzing its code. I'll leave the thread open for the community to continue the conversation.

See you all in Tamriel, and have a great time gaming! 🫶

You need a better search engine or chatbot because it is providing you the wrong information.

tomofhyrule · June 17

One thing to remember is that the "lol just replace the engine bro" argument has been made several times in the past, and it is just as unrealistic each time. Kevin even directly responded in some of those threads as to why that is not possible:

ZOS_Kevin wrote: »

As a few people have mentioned, that's not how game engines work. It's not like you can copy and pasting code into Unreal Engine 5 and therefore ESO will run in Unreal. That ask is making an entirely new game. There are things that we can do like the next-gen overhaul we did for consoles a few years ago and work within the existing engine for some graphical improvements.

ZOS_Kevin wrote: »

Just wanted to chime in here. Yeah an engine swap is not likely. That literally requires building ESO as a new game from the ground up. So that likely is not going to be the case. Halo going to Unreal does not mean that Halo Infinite is being ported to Unreal. It means that future Halo titles will be made using Unreal. However, that does not mean there are not things we can do to enhance visual quality with our current engine. We have passed this feedback to the team from various threads to see how we can tackle this problem in-game. So the team is aware.

Additionally, the engine is not responsible of content output. Every engine has things it can and cannot do. However content wise, our current cadence as some have already mentioned, was adjusting to player feedback from a few years ago. However we are hearing different feedback now and have passed that along to the powers that be.

If the argument is instead that server hardware needs to be replaced, that could be more reasonable of an ask... if it weren't 2026 and hardware prices weren't exorbitant. With Microsoft demanding a 30% ROI from all of their subsidiaries, there will need to be some solid proof that new hardware would bring in a lot of extra money to justify the cost, not just "but Cyrodiil is laggy."
As a sidenote, people have said that server hardware usually has a 3-5 year lifespan. But we could also talk about the lifespan of the client-side hardware as well, aka older PCs and last-gen consoles. Even brand new servers would still get throttled if those components are insufficient as well.

ESO_player123 · June 17

Gabriel_H wrote: »

VidmaVirtual wrote: »

Gabriel_H wrote: »

VidmaVirtual wrote: »

Your argument that player actions must be entirely sequential on a single thread to prevent race conditions is based on legacy, object-oriented MMO architecture. Modern next-gen MMO backends solve this challenge using Deterministic Lockstep, Entity Component Systems (ECS), and Spatial Partitioning.

That isn't my argument. My argument is that player actions ARE sequential. That has to run through a single thread. I'm trying to keep this simple as you can't even grasp the difference between engine, game code and server.

So lets complicate it up.

VidmaVirtual wrote: »

Modern next-gen MMO backends solve this challenge using Deterministic Lockstep, Entity Component Systems (ECS), and Spatial Partitioning

Deterministic Lockstep: Is client A to client B, rather than Client A to Server to Client B. The latency of that is determined by the player with the HIGHEST latency. So someone in Australia with a poor ISP running at 500ms latency means EVERYONE has high latency, freezes and delays.

Entity Component Systems: This could boost performance, but the bottleneck remains the calculations. Separating the object to asset & data doesn't change the number of calculations on that data.

Spatial Partitioning: ESO already has this. That's what instances are, and what phasing is. Doing it on the fly in a map the size of Cyrodiil would require splitting the map into multiple grids. With so many dynamic rather than static components this would add not reduce the load on the servers.

@Gabriel_H
If you are going to condescendingly tell me I "cannot grasp" technical concepts, you should at least make sure your own definitions are accurate, because almost everything you just wrote is fundamentally wrong.

1. Deterministic Lockstep: You are completely confusing modern server-authoritative determinism with 1990s Peer-to-Peer (P2P) network code used in old RTS games like Starcraft 1. Modern server determinism does NOT slow down the entire simulation for a high-ping player in Australia; the server dictates the master state tick, processes the parallelized inputs, and rolls back or drops late packages without affecting other players.

2. ECS (Entity Component System): You claim separating data from logic doesn't change the number of calculations. True, the *number* of calculations stays the same, but *how* they are executed changes completely. Instead of a single master thread processing heavy player objects one-by-one in a long slow line (sequential), an ECS arranges data linearly in memory cache, allowing modern multi-core server CPUs to process those calculations simultaneously (parallel computing). 32 cores doing 1000 calculations at the same time is lightyears faster than 1 core doing 32,000 calculations sequentially. That is how bottlenecks are broken.

3. Spatial Partitioning: Phasing and instancing are NOT spatial partitioning. Phasing puts players in entirely different visual realities. True Dynamic Spatial Partitioning (like server meshing used in modern network infrastructure) dynamically chops a live zone into virtual grid cells managed by different hardware nodes on the fly. It doesn't add load—it drastically reduces it by isolating the combat calculations of a castle siege so it doesn't lag out the rest of the map.

But as I promised myself before writing this: this will be my final response in this thread.

Thank you so much to everyone who jumped into this discussion over the last few days—including you, Gabriel_H, for keeping the debate lively. I really appreciate all the deep technical breakdowns, the macroeconomic insights, and the different perspectives we shared.

I have fully laid out my views as an 11-year veteran of ESO. Since these forum debates can easily spin in circles forever, I am stepping back to actually enjoy playing the game instead of analyzing its code. I'll leave the thread open for the community to continue the conversation.

See you all in Tamriel, and have a great time gaming! 🫶

You need a better search engine or chatbot because it is providing you the wrong information.

I understand that OP is passionate about the topic, but I would really prefer worse translation to this type of writing. I just cannot bring myself to read that AI style.

Blood_again · June 17

"Using as a translator" is a usual way how LLMs explain chatbotting by them.
It is far from truth.

Giving 4 examples of game engine updates with technical details, that took 6 minutes summary: to read the question, find and compose examples, add and check technical details, finally translate.

Who, you think, has generated this whole answer: a human with their text + machine translator, or the LLM + copy-pasted question + a simple prompt?
Using Occam’s razor would be a plus.

spartaxoxo · June 17

The devs will do it when it makes financial sense to do it. As long as older consoles bring in way more revenue than it would be to cut them out, the devs will make adjust the game to accommodate them as long as it's technologically viable.

They may even go the sequel route and leave this game to the OG consoles. Some games do that when redoing the whole game and losing that revenue costs more than it is worth.

Ultimately, whether or not old consoles hold things back isn't just a matter of technical compromises being made but also how much of the pie they make up financially.

valenwood_vegan · June 17

They just went through layoffs and restructuring and a leadership change and haven't even proven their new model yet.

I doubt anyone is suddenly about to give them the necessary funding and time for a costly multi-year overhaul of the whole game, particularly in the current environment. It's worth noting that ZoS currently isn't even hiring *at all*.

Hopefully they continue to move in the right direction and improve what they can within their limitations, win players back and attract new players, show that the seasonal model is bringing in players and increasing income, and get ESO into a good place where things like a major technical update or new hardware are on the table, which would of course have to be backed by more than forum posts and wishful thinking: it would need a strong business proposal with specific data, timelines, costs, financial projections, etc.

freespirit · June 17

Just stop trusting AI generated information or maybe just as an experiment ask multiple different AI's the same question, I absolutely guarantee you will get multiple different answers!

The wall's of text are bad enough but inaccurate wall's of text are just annoying.

As mentioned above Kevin has already explained why what the op is requesting wont happen.:(

c363b · June 17

This thread could benefit from a comment from an ESO developer.

wolfie1.0. · June 18

VidmaVirtual wrote: »

wolfie1.0. wrote: »

VidmaVirtual wrote: »

@Aislinna

That is not how business accounting works. You cannot amortize active live-service revenue over pre-launch development years when the product wasn't even on the market generating sales.

The initial development costs from 2007 to 2014 were capital investments that were cleared and paid off a long time ago within the game's first few successful years.

We are discussing the current financial health of the game in 2026. Evaluating a live-service title's ability to fund a modern upgrade by dragging in pre-launch sunk costs from nearly two decades ago is a massive reach. The reality remains unchanged: the game generates immense monthly revenue right now, and reinvesting in its core technology is standard product lifecycle management.

Yes, lets talk about the curent financial health of ZOS.

You have to realize some realities here. You picked the absolute worst time to purpose a full rebuild. ZOS does not operate in a bubble.

First of all from what i understand based on my limited knowledge of GAAP you can Amortize developement expenses over the course of the anticipated lifetime of software once its launched to the public. So the cost wouldn't start really decreasing until 2014. But the real question isn't how well ESO and ZOS HAVE performed, its how they are doing now. So past earnings and figures prior to 2021 honestly dont matter. Once the Microsoft aquisition occurred how well ESO and ZOS preformed prior to then only matter in the sense that Microsoft was sold on an expected rate of performance and Return on Investment.

So lets talk corporate. $7.5 billion USD, that is the important number. That is what Microsoft paid to Zenimax private owners to aquire Zenimax, its publishers and game studios. It finalized in 2021, and folded it under the Xbox segment of the corporation. There is no denying that Microsoft could afford it, and could afford to take a loss on it, but corporate and the shareholders would love to see that investment provide an solid return.

So, lets look at the state of Xbox... its a mess, i wont bring up the details here but the news is everywhere. Xbox is NOT doing well. Projects, Studios, and games are folding. Consoles are NOT selling. Gamepass has gone through several pricing changes. Leaderhip has changed. Microsoft has redirected to AI and has invested unreal amounts of money in it.

Meanwhile Xbox is under new leadership, and while not publicy stated the implication is pretty obvious, turn a profit or be gone. New leaders have stated that they are going to invest in core IPs and properties. Thankfully ESO is technically a core IP, but its not mainline Elderscrolls, that could be a boon, it could be a problem, or it could be status qou. Keep in mind that ZOS was forced to cancel a project for a new game that was estimated to have already invested around $400 million (if not more) and created a leadership change. ZOS still needs to recover those funds.

All of this is happening in an enviroment where silicon costs have risen to extreme levels due to AI data center demands. RAM, GPUs, and Storage costs have increased substantially. RAM and storage 2 years ago would have been the cheapest components of any build (including consoles). We have seen price increases on EVERY major system for tech that is 4-5 years old. If Stadia launched today, it might actually have had a better chance to live on longer than it did.

Its in THIS enviroment that you want them to update the core code. essentially rebuild the game from almost scratch, and then relaunch the game, to what would be a smaller audience. Its simply not going to happen, at least not yet.

Honestly, ESO will likely be held to together as best it can until revenue catches up with the changes and recent upheavals in the industry. The Focus for Xbox is to get both a new Elderscrolls game, and likely a new Fallout game launched first. Then IF those are successful enough and IF Xbox shows a profit, then and only then do I see a possibility of getting a new engine and full build of the game.

until then, i think we should be a tad grateful that we still have a game to play and devs that are actually trying to improve the system. Some other live service game communities are not so lucky.

@wolfie1.0.
Thank you for this incredibly comprehensive and highly accurate macroeconomic breakdown. You are entirely correct about the current corporate climate, and I really appreciate you bringing these hard realities into the discussion.

The Xbox segment is indeed facing massive pressure, the rising hardware/silicon costs due to the AI boom are real, and ZOS cancelling a $400+ million project undoubtedly created a massive financial wound that the studio is desperately trying to heal. You are 100% right that from a traditional corporate standpoint, picking this exact moment to pitch a high-risk, expensive engine rebuild looks impossible.

However, if we look at it from the perspective of Microsoft’s expected Return on Investment (ROI), there is a counter-argument to consider: the cost of letting a core premium asset slowly suffocate.

As you mentioned, ESO is a core IP property that brings in highly reliable monthly recurring revenue. But that revenue relies entirely on player retention. When long-time players, high-end trial groups, and massive PvP guilds gradually walk away because the technical stagnation makes prime-time gameplay unplayable, Microsoft's expected rate of performance begins to decay.

Neglecting the core infrastructure to save short-term capital is how massive live-service games enter a terminal downward spiral. If the engine is currently so bottlenecked that ZOS has to severely slash campaign population limits just to keep the servers from crashing, they are fighting an uphill battle against their own product's expiration date.

While I completely agree with you that Xbox leadership is being hyper-reactive and focusing heavily on the upcoming mainline Elder Scrolls and Fallout titles, allowing their most reliable, active MMO to slowly decline due to technical debt is a dangerous strategy.

You made a brilliant point that we should be grateful to still have a live game with active developers—and I truly am. I love this game, which is why I’ve supported it for 11 years. But it’s precisely because I want to see it survive another decade under Microsoft's messy corporate umbrella that I believe a long-term architectural solution needs to be on the table, even if the current industry climate makes it incredibly difficult. Thank you for such an amazing contribution to this thread!

I appreciate your counterpoints and understand them. I do think your still missing the point i am trying to make, but you did mention it if not directly.

Microsoft, Xbox, and ZOS are in a state of consolidation and financial strain. They will not invest in a high endeavor right now. Yes, there is a risk of losing ESO player retention and people leaving the game. It IS a problem and a concern.

However, Rebuilding the game from scratch with new graphics wont fix that. Plenty of games have fancy graphics and can't seem to attract enough players to survive long. you need a draw that is more than just looks. There is also no gaurantee that a new build wont have the same issues as the current one.

then there is the factor of time. Even if they started development of ESO 2.0 today you are talking 4-6 years worth of work to get it online, probably longer as you will have employees or new staff needing to either learn about ESO or learning the new engine.

meanwhile, the underlying issues might be resolved with 2-3 years of developement under the current system.

I wont even get into the financial benefits Microst gets from not needing to retain employees or servers.

One of the reasons ESO is in danger is because its a live service game. Skyrim has very likely made as much or more than ESO over both products respective lifetimes and Skyrim costs significantly less to maintain. which means that a successful sequel could provide a significant moentary boost. A new IP wont have that kind of performance, and a live service game has ongoing expenses.

i look at it this way. Lets say you own a home, and have owned it for 30 years. its not the best home in the world, not your dream home but it serves your needs and does it job, if not reliably well. Taxes are increassing on you and you find that you need to do a bunch more stuff overtime to keep it in a way that works for you. This morning you find out that your roof is leaking and there are several locations that need to be replaced. Its not covered by any insurance policy though, so any repairs are fully on you to make.

Now the reasonable approach is that you either patch those specific parts of the roof in the hope that you can get a few more months or years out the roof, or you may just bite the bullet and replace the entire roof so you can extend the life of your home. What would be insane to do would be to demolish the home and rebuild it from the ground up, especially if you dont have the money to do it.

What is happening here is that the roof has been damaged for ESO. It NEEDS to be replaced. Patches are limping that need along while the roof is being prepped for replacement. What you are asking instead is for ZOS to destroy the house and rebuild it from scratch instead, when there ISNT the money for it.

SkillzMFG · June 18

SneaK wrote: »

I cannot see how anyone would argue against OPs suggestion. Like why on earth would we not want ZOS to upgrade their tech?

People don't want to 'take up the devs time' for seemingly...any reason? From Cyrodiil, PvP balance to PvE Balance and trials, and combat in general.

Just stock up that Crown Store with more 30€ cosmetics I guess? Who cares if the game is good, or God forbid even works properly.

SneaK · June 18

SkillzMFG wrote: »

SneaK wrote: »

I cannot see how anyone would argue against OPs suggestion. Like why on earth would we not want ZOS to upgrade their tech?

People don't want to 'take up the devs time' for seemingly...any reason? From Cyrodiil, PvP balance to PvE Balance and trials, and combat in general.

Just stock up that Crown Store with more 30€ cosmetics I guess? Who cares if the game is good, or God forbid even works properly.

Right.. I can’t pretend to have an ounce of know-how related to most of what OP is saying, but it seems as though to be a thought out suggestion. Which, if any suggestion would result in a net positive why are we arguing? Forum has always had white knights but it’s worse now than ever.

Pepegrillos · June 18

SneaK wrote: »

SkillzMFG wrote: »

SneaK wrote: »

I cannot see how anyone would argue against OPs suggestion. Like why on earth would we not want ZOS to upgrade their tech?

People don't want to 'take up the devs time' for seemingly...any reason? From Cyrodiil, PvP balance to PvE Balance and trials, and combat in general.

Just stock up that Crown Store with more 30€ cosmetics I guess? Who cares if the game is good, or God forbid even works properly.

Right.. I can’t pretend to have an ounce of know-how related to most of what OP is saying, but it seems as though to be a thought out suggestion. Which, if any suggestion would result in a net positive why are we arguing? Forum has always had white knights but it’s worse now than ever.

OP is a an LLM parrot and all of you are getting baited.