Playerunknown’s Battleground was pushed back to an unannounced date in 2017. Assuming it’s the late evening of the last day of the last month, December the 31st 2017, and assuming two weeks prior of feature locks (or Release Candidate, basically just closing opened bug ticket and adding nothing more to avoid side effects), that leaves BlueHole three months to get out of the Early Access “we’re asking for your patience” theme into a truly published game.
This post isn’t a rant. Won’t be talking about microtransactions and business shenanigans. Won’t be talking about general bug fixing (and the hundreds of opened tickets acknowledged by BlueHole
). Neither will it be about cheating, as it’s an ongoing effort. Nor will it be about pet features added, either mine or yours. It's fully about what BlueHole said, or it's clearly mark as a personal guess. The TL;DR version is: PUBG is three months away from release. BlueHole have huge difficult chunks of work to do just right in that time, according to themselves: User interface complete rework, Training area/Shooting range, Vehicle physics rework, Bullet physics rework, Benchmarking tool, Audio mixing and volume issues, Client optimizations, Network code and predictions, Servers stability and performance, 2 new maps, Melee takedowns, Vaulting. That’s at least 12 very big challenges, on top of everything else, to do in only three months. That’s at the most three patches.
User Interface rework
A complete rework of the user Interface was announced as “in the work” very early. It’s not only about graphics, although a 2 meter long AKM in the lobby clearly has some issues. It’s first and foremost about user experience design, and probably include features and bugs relating to creating and managing duos & squads, navigating leaderboards, advanced configuration panels, adapting in-game user interface for various play mode (such as first person and third person), accommodating various hardware such as ultra wide screens, rephrasing a lot of the current options that do not do what they say or are counter intuitive, and so on.
In particular, one thing is curious: as far as I know, there’s absolutely no exchange with the community. No alpha or beta versions of the new user interface. This is one thing that’s usually the golden child of fast and numerous iterations in the industry: it takes a lot of steps and lots of tries to get it right. It could be done with large dedicated teams with a ton of experience, and the budget to do multiple focus groups worldwide and so on. But I doubt that’s the case. And a bit of a waste.
Training area & shooting range
A very very old feature request, as early as January I think, and an extremely common one. The game has a slow iteration time (not like Overwatch and the like): matching, loading the island, waiting for other player, then long countdown, then plane ride, then parachuting, then landing. And now, if you want to test something, well you can’t. You have to survive, and find it, and then waste part or all of your round to do some light experiment. Shooting 100 bullets isn’t training, it’s a warm up. They acknowledged it, and promised to implement it later during Early Access, judging it less important than the first patches.
It shouldn’t be that hard to do. It’s a very small map with assets and items in it, and maybe some extremely basic bots as targets. It can be self-hosted to accommodate the online nature of PUBG client, much like I would guess modders will work in the future.
Clearly, by design or by accident, the vehicles part of the current game lean heavily on the GTA sides of things. But even in that paradigm, we’ve all seen multiple times in our games, or at the very least in video highlights, the bizarre physic of vehicles. Jumping 5 meters for a 20 centimeters bump, doing 3 barrel rolls for a light bush, getting thrown dozens or hundreds of meters away for a collision, or not budging at all while been rammed by a 100km/h UAZ, people or other vehicles inserted in your own vehicle, and so on.
This is not an easy thing to do, because there are several things in play. Models, hitboxes, netcode, server and client issues, and so on. It’s a big chunk of work, and since Early Access in March we’ve seen little work on it. A few less bugs, but nothing major changed. Some heavy, heavy work awaits a team with experienced people in it.
That’s one thing I haven’t seen that much, and don’t understand why when literally ten of thousands of message talk about client and graphic performance and can I run this, and how to gain 3FPS here or 2 and a half FPS there. It’s probably such a basic, self evident feature, that everyone (media included) assume it would be made available sometimes later during Early Access.
I do find this missed opportunity strange though. Because BlueHole did do some work these past 7 months of Early Access, including optimizations, and a stable benchmark tool would have been helpful in a PR perspective to show the amount of work done. And more importantly, I would imagine it would help a lot the developers and the QA people when reading user post about FPS drops since this patch or that patch, to have clear metrics through a benchmark tool.
It’s not a lot of work to implement a good benchmark tool: determine the % of play in this or that aspect of gameplay (red bombing zone, light forest, on the water, in the water, in a town, in a slow car, in a fast car, etc.) and macro a rendering of prerecorded real game experience with those percentages. Record a lot of metrics, levels, and plot nice curves. And I’m sure they have those tools internally. But, it’s still work to implement for the general public.
Another one old announcement, from early May if I remember correctly, about the bullet physic being fully reworked. It’s not a huge chunk of work per se, although having something heavily optimized still takes time. But it’s something that create work: Ok you’ve re-done your physics, great. Now we got to do another balancing pass on everything, make sure nothing was broken, make sure no new exploit can be made. And so on.
We’ll all experience it, the current game has audio issues. Yes there are bugs, saturation issues for multiple sources for some, and so on. But there’s the current state of the mix volume being uncomfortable at best, possibly hear damaging at worst. Part of it is a desire to be grounded in verisimilitude, which is admirable. But anyone how rode in a 70’s car knows very well none of them were that loud ;) And of course, no verisimilitude can’t damage players health
BlueHole stated that nothing is definitive, implying they are working on this. And 22 millions ears surely hope so. It’s not a lot of work to do a basic pass on the audio mix, it’s literally either some numbers in a configuration file, or some blob to move down when exporting the audio file. And it can be done by ear, just an approximation, then iterate it up and down. It do feel strange the last seven months of Early Access weren’t used for those iterations, especially since it has close to no bearing on what the other developers do in a patch.
A more advanced solution was given as early as May on the official forums: earplug, Arma style. Again not a lot of work to give every player an item to put in ears that block let’s say 80% of sounds, and can be remove at any time. The most work is animating putting it in and removing it. It won’t solve all of it by itself, by with a more sane audio mix as described above, it could be a small feature to accommodate a more Military Simulation verisimilitude angle on the game, while keeping it playable.
Do I need to say more?
Network code, client and server predictions, tempo management, etc.
We’ll all felt it in various ways. We’ve seen ten if not hundreds of thousands of message on it, everywhere. After the graphic client optimization, it’s probably the most hot issue. It concern the general lag of taping a button and waiting way too long for something to happen, but also hit registration, vehicle collision, and the abuse of the system. The current network code has some peculiarities, to keep this simple it’s extremely forgiving of high latency (“ping”) players, to the point of sometimes being at an advantage against a low latency player. In the industry any network developer knows there’s no Holy Grail about this, it’s all about balance and compromise.
There has been some work about this, and we’ve seen visible improvements late spring. Not nearly enough, as was acknowledge by BlueHole, who’s still working I would guess on both optimization of the code, and changes to the balance between high and low latency handling. But usually this is a highly iterative work, and hard to do in a small studios because real world conditions (network condition, but also player psychology, hardware, and so on) varies considerably. It’s a lot of work, and it takes experienced and talented engineers, and usually a lot of time and iterations to get it right with the proper community metrics and feedback.
Server stability and performances
Another hot potato, and a separate big chunk of work from the network code above. This would include anything related to the authentication servers (the probable thing that’s closing your connection when you want to play, give you an empty inventory from time to time, maybe mishandle your BP and your player crates), but also to the brute performance of the game server (like having an average tick rate of 15-20), and its stability.
It’s a serious chunk of work, probably not requiring the best engineers around, but a lot of legwork to look at things packet by packet, tick by tick, hundreds of hours of profiling every aspect of the server software to see where are the bottlenecks, a lot of log reading to track and reproducing crashing bugs, and so on. Apart from acknowledging the huge amount of work needed here, BlueHole doesn’t communicate much on this (and the little we got was contradictory or inconsistent at times), and most patch notes are vague or empty about it, so we know very little apart from what’s happening in our games. Clear visible improvements were made, from the depth of unplayable some days to just buggy. But still a lot of work to do.
We know two new maps are in the works for Early Access, with more details on the desert one that the varied biomes or mountainous one (whatever you want to call it). Both a huge chunks of work. Huge. But with a big advantage, both are doable by new people, without interfering with the work of the core team. To put it another way, new map work can be done in parallel of the main optimization, bug fixing, and so on.
That’s also a highly iterative work, although usually a big part of it is done internally. Fortunately, we have Erangel (the current map) to see the time it takes. Even after 7 months of Early Access, there’s still months old tickets opened for bugs on Erangel. Small things at that point, holes in the ground texture, small part of the map one can fall through, some objects that should be destructible and aren’t, that sort of things. And just this month of September, we saw a new iteration of Erangel, with a new town added to balance it better, as per a lot of community feedback dating back at least from February or March. But that gives an idea of the time it takes to do things.
Sometimes perceived as a small feature by some player, it certainly isn’t. It will have far reaching implication in the general gameplay (for the better most hope), and is some serious work: a lot of animation of course, but also coding, at a lot of iterative work to get it “just feeling right” and balanced. It’s something that’s close to impossible to balance properly internally with a few dozens people, millions of players will find way to break it. It does need some serious open beta testing to get it right.
And last but not least, the BIG ONE. Vaulting is a huge, huge amount of work, from a lot of different departments. It requires a lot of animation, a lot of coding, coding, user interface, level design input, assets creators and managers inputs and probably work, and so on. It’s a personal guess, but I think the success of PUBG made this one possible. It was very probably in the “Pipe Dream drawer”, and after a few millions copies sold, was dusted off late spring. Taking it months to implement, when it will probably have so much impact on gameplay and player tactics and experience of the game, is quite understandable. It’s to the credit of BlueHole to have limited feature creep during Early Access, this been one of the only two I saw (basic things forgotten before like custom binds or toggling or such aren’t feature creep in my book, they’re bug fix), with Takedowns.
But details are what make or break such a system. Does it feel right? Is it responsive in the right way? Does it work good enough in third person, first person? A lot of work is done internally, but experience shows only a big amount of real players can yeah or nay it. Usually if 90% of the work can be done internally, the last 10% has to be iterative with player feedback. And it takes times, especially when the game is already played and some time is needed for the players to ease into the new system.
For something so integral to the continued success of the game (according to BlueHole itself), there’s a deafening silence on the subject. Hence, it wasn’t put with the other.
Conclusion and thoughts
I wonder what’s the community think of the challenges ahead? Given the amount and rate of work that BlueHole has done since Early Access in March, 6 months ago, what can we expect at general release of PUBG?
Do you see three whole big patches until the end of the year, each with 4 of these big chunks above fully fledged and done, on top of the general bug fixing?
I was reminded two things:
was in fact promised before leaving Early Access. To quote the official website “Upon launch, the Battle Royale game mode will be complete, and we’ll have implemented full modding support.”
Since technically Benchmarking wasn't as far as I know announced, we can switch those two in the listing.
Second, I also (clearly, I need to buy a new brain on eBay) forgot the Replays, Observer mode, Kill cam
, and everything around that topic. So we are now at 13 Labors of Hercules, plus one.
New edit, I also forgot the ELO system. But that's such a small amount of work compared to the other, I wouldn't include it in the 13 Labors of Hercules. But I may be wrong.