Sunday, April 10, 2016

[WoW] Classic Blizzard-Run Servers? Code, Logistics, Marketshare Point To "No"

Wilhelm over at TAGN had a great post the other day on Blizzard's shutdown of the Nostalrius Classic WoW private servers, talking about the potential market and reasons why or why Blizzard might not go the route of creating their own Classic WoW server.

Personally I'm of the opinion that no, we won't see Blizzard-run Classic WoW servers anytime soon. Mostly because of code, logistics and possibly not enough profit to make the investment risk worthwhile.

" get the old hardware, the old code, because the old code is meant to run on the old hardware, the old data, the old bugs, all that kind of stuff. Of course the natural expectation is that well you would fix all that stuff." -- Tom Chilton
The biggest barrier for Blizzard's entry is probably just getting old code up and running. As Tom Chilton's quote above indicates, assuming they could get the old code out of their backups--depending on how their code repository stores 10 - 14 year old data, which is a huge potential issue by itself, especially around art assets--they also need to rebuild the old hardware for the server code. There'll be code that relies on timing or performance characteristics of CPUs, RAM, internal networking, etc.

And if they couldn't get the old hardware again, they'd have to fix any number of potential bugs that would be caused by moving to a new hardware profile. Trying to figure out if that timing issue is a bug that existed in Vanilla, or due to hardware modifications.

That also doesn't take into account that their server OS--likely a Linux or UNIX variant--has had 10+ years of security, performance, and API tweaks. They certainly don't want to use a 10 year old OS for security issues, but the code may not even compile correctly anymore because OS modules have evolved over the years. Heck, for both client and server, they may need to be adjusted for newer compilers in general. Not to mention that Blizzard's current build pipe has evolved such that it would take more work to adjust Vanilla WoW development to fit the build pipe.

Speaking of security, WoW servers would have had a number of bug fixes over the years for security and anti-cheating technology that would be wholly missing from Vanilla WoW. Blizzard certainly wouldn't want to ship security holes even if they decided the anti-cheating tech wasn't worth the effort, just because it could potentially leave the rest of their network compromised. Those bug fixes would have to be identified and ported back.

Also, Blizzard's Authentication servers have evolved over the years, including support for 2-Factor Auth and likely protocol changes to the auth service itself for security reasons. Those would have to be back-ported into Vanilla WoW.

Then there's also the client itself, which would possibly need tweaks to handle newer graphics cards. Theoretically DX11 and DX12 are both backwards compatible with DX9, but that's not to say there aren't graphics card specific issues. Even on Eon Altar for Unity we've hit the occasional graphics card that just barfs on things and needs a specific solution. The cost here is almost entirely on the test team rather than the engineering team, but it's still not cheap.

There's no Battle.NET integration in Vanilla WoW on the client or the server, so that's another feature they'd have to port, and that one's a doozy. Part of it likely would come with the auth server changes (since they hook up with BNet), but current friend lists across games would need to be re-implemented.

It also ignores any further bug fixes to the game. These might include networking optimizations to make the game more responsive/efficient, content issues, systems bugs, and so on. While folks might be okay with Blizzard shipping a buggy game as is for nostalgia value, Blizzard's quality bar internally is probably set higher than that.

None of the above is impossible. Just an immense amount of work, and not all of it easily identified, especially in the cases of security and hardware bugs.

"But kind of maintaining that many different versions of the game is just not really feasible. Particularly in a world where people that are playing right now really want more content, not less." -- Tom Chilton
Let's say we've identified all the potential code issues and now it's a matter of assigning people to perform the work. If we pretend that five programmers are sufficient--say, 1 senior lead, and a junior plus mid-level programmer pair for both client and server--for a year, you're talking about $600,000 to $750,000 for salary, benefits, HR, legal, equipment, and so on.

That also doesn’t include testers, build teams, deployment teams, server hardware, server ops people, data center hosting costs, marketing, and more I'm likely missing. Testing alone would be a massive endeavor, and a lot of the testing would have to be extremely technical in nature given the hardware and security issues we've potentially identified.

All of those people could be working on the next Hearthstone or Overwatch instead, so there's an opportunity cost here that's harder to quantify. Or, even working on more current WoW content as Tom Chilton mentions above. Splitting their development team when they can barely put out content fast enough as is doesn't seem wise.

I'll ballpark a figure of $2M over the course of a year for this project, though I may be undervaluing it significantly. I don't have good figures on how much marketing, testing, or data center hosting costs. Suffice to say, MMOs are expensive, even if you're starting with an existing code base.

Profit vs. Risk

If $2M is the price to beat, then Blizzard would have to sell ~133k subscription months to break even at $15/month, and that's if we ignore taxes. If I ballpark a ~16% corporate tax rate from these investor values, they're actually looking closer to ~155k subscription months to break even on the initial layout. That doesn't take into account operating costs for support, test, devops, community managers, game masters, data center hosting costs after the initial deployment, further marketing, and so on.

Nostralius was free, but they claimed to have 800,000 registered users and 150k active users on their server. Alyson Reeves used micro-transactions to net a cool $3M from her private server before she got shut down hard. So there is clearly money to be made, but the question is, is it enough?

If we assume a 100% retention rate for Nostralius customers transferring to Blizzard--which is ridiculous at face value--then Blizzard could likely break even, and make a little profit potentially.

It's not really an apples-to-apples comparison, mind, because Nostralius was in a gray area at best, and a Blizzard run server could garner customers uncomfortable with gray or black market activities, similar to how Blizzard did the same with gold buyers and the WoW Token. But it's also not a fair comparison because it's highly doubtful all of those people would pay $15 a month to play Vanilla WoW again. Similar to how RIAA claiming that 100% of pirated music count as lost sales is spurious--many of those people wouldn't have paid for the music regardless.

As an extremely rough guess, if we assume that a Vanilla server probably would mostly get 1 month tourists who wanted to just get that warm fuzzy nostalgia feeling, then if we pretend that 650k of those Nostralius users were 1 monthers and the 150k active users were 6 monthers, that turns out to be about 1,550k subscriber months. Or 1,302k after 16% tax, or 1,169k after the initial layout is subtracted. $17.5M is nothing to sneeze at if subscriber months netted $15 each (without taking into account operating costs).

But it's also a dead-end. They can't monetize Vanilla WoW the same way they can monetize Hearthstone, Heroes, or WoW. They can't release new content for it the same way WoW expansions work today, short of repeating the process for TBC, Wrath, etc. Also, no WoW Tokens, no cash shop, and do we really think Vanilla subs will actually pay $15 a month for an old product?

So yeah, they could grab a decent chunk of change for a one-time fee, but it's not a slam dunk and it's not actually that much as far as a Blizzard investment is concerned. It'd also be interesting to see if they could actually get enough employees interested in such an endeavor that they'd choose it over a newer project with more opportunities, and as mentioned in the previous section, the opportunity cost and potential of taking these employees who could be building new stuff could be lost profit as well.

Signs Point To No

There's a lot of things to keep in mind, and I'm sure there's a lot things I missed.

Assuming Blizzard could actually get the correct code, get the hardware, and get everything set up for the modern Internet and modern systems, they could probably make a neat profit. But because there's little to no room for growth after that layout, I don't see them making a huge investment here. It just doesn't seem worth it in the grand scheme of things.

At the end of the day, Blizzard--and all video game companies--are still a business.

#Blizzard, #WoW, #GameDevelopment


  1. Pretty much sums it up. Call me cynical but I don't think it would be "enough." Eventually the people who wanted the Vanilla server will get their best-in-slot from Naxx/AQ40 and run out of things to do. Then they either quit or start begging for new content (or to have content from other expansions somehow bumped over). Some might roll another character, some might stand on the cart in Ironforge, but eventually they run out of things to do.

    I am curious how Nost and others solved the hardware issue. I did not play on it but I wonder if they encountered issues regarding old code being on newer hardware; or if they sprang for the old hardware to run it.

    1. For private servers, they're not running Blizzard's server code. They'd have written their own code base reverse engineered from how the client interacts with the server. A technically impressive feat, but comes with the issue that it'd be buggy because some guesswork would be involved.

      Blizzard resurrecting old code means it has requirements to run the old code. A private server written from scratch has no such restrictions.

  2. As someone who would love to play on a vanilla server, but refuses to torrent, I think there are three key issues with the common anti-vanilla argument.

    1. Nostalgia makes me want to play a bad game.
    I liked Vanilla. I actually prefer TBC, but I never raided during Vanilla, so there's that appeal to me. I prefer the game design of that time period. Maybe not all of it, but enough to put up with the stuff I didn't like.

    I would suggest that the majority of people who enjoyed that design have long left WoW (I left 2 years ago despite having prepaid for Warlords). So it's not surprising to see most active WoW players talking about "how bad the game was back then".

    I don't like slurpees, but I don't go around telling people who do that they're wrong/crazy and only think they like slurpees because they did as a child.

    2. Blizzard would be better off using those resources for current content.
    To me this is a variation on the old strawman "would you rather have X feature than a raid tier". I think at this point we can all just accept that they are terrible at producing content in a timely manner for the live game. Warlords only had two raid tiers and they still have about a year of dead time planned.

    3. It would be too hard to do.
    I suppose this makes sense in the same way that a company that pulls in about a billion dollars a year in subscription funds can't seem to figure out how to release new content in a timely manner. Well, new content for WoW. They sure did use that money to produce three entirely new games and a sequel (Hearthstone, HotS, Overwatch, and Diablo 3).

    I release that's a flip comment, but there have been requests for Vanilla Servers since Cataclysm at least (The Ancient Gaming Noob has some quotes), and Blizz has had more than enough time/money to get this done if they wanted to.

    Even now, it's probably safe to assume that Blizz has at least a million active western subscribers. So they have $15million per month from WoW that they certainly aren't spending on the live game in a noticeable way. So they certainly have the funds/option. They just don't really want to.

  3. I feel like you argued that people use a straw man argument regarding vanilla servers potentially taking away development resources towards live, and then went on to make your own straw man argument suggesting every other Blizzard game took away development resources from WoW.

    And are these arguments against legacy servers, or explanations for the actual hurdles that would exist? Is he really saying they shouldn't do it? I see it as discussion that should drive better arguments. If you understand the issues more clearly, you can focus on other reasons for why they should do this. Reasons that actually makes sense.

    It doesn't have to be, "Well, Nost did it so you could do it easily!" We don't fully understand what Nost had to do (or Mangos) or what Blizzard would have to do release this game to their own standards.

    Instead the arugment could look like this: "Well, you guys have always supported your old games (D2, SC, and WC3), and it is one of the things your fans loves about you. Initial monetary benefits might be risky at first, but it would be a great PR move in a video game publishing environment that screws over customers with shady monetization methods."

    They have a culture and design philosophy that lines up with supporting their games, and I'm sure they want to preserve their own franchise's history. You can appeal to that and argue for why you love the old game want to play it again. That's really it, no need to make up an 'easy' solution without understanding why it is in fact not easy. That's what many people in the pro-legacy camp resort to, and it is unproductive.