Jump to content

Fireside Chats (with a video game designer) - July 10, 2014

Recommended Posts

  • Developer

Our First Playtest


As I mentioned in my last post, we had our first office-wide playtest of Don’t Starve Together on Friday, July 4. Okay, that’s not entirely accurate: we’d done one a few months ago as a proof-of-concept. This was our first playtest after Don’t Starve Together had fully gone into development. It was the first time we were playing something that wasn’t simply Don’t Starve synced across a network with automatic respawning (and a bunch of problematic stuff ripped out). The first time we were truly playing Don’t Starve Together. Approximately 15 people in the office were able to spare some time to participate in the playtest, which lasted 30 minutes or so. Overall, it went well. Inevitably, we discovered a handful of bugs that we’ll have to address. None of the bugs were showstoppers, though: people were able to play and we learned a lot about what was working well and what we would need to iterate on. More on all of that below. As before, I'm going to spoiler tag the meaty (i.e. anything besides the intro and closing remarks) sections, because I apparently have no idea what brevity is.



Some Footage

As the title card on the below video says, this is early footage that shows some prototype mechanics we have roughed in--they’re still being worked on. It’s by no means final and everything you see is subject to change (though, as we’ve said, we don’t have plans to make drastic changes to the core of the game). In light of that, don’t think of this as a trailer--instead, think of it more as a glimpse of what we’re experimenting with right now. Specifically, I wanted to show you a game running with multiple players in it, the first version of chat between those players, the first version of the ghost mechanic, and how resurrection works with that ghost mechanic. Without further ado, some Don’t Starve Together early gameplay:


Let me reiterate: this is early footage of prototype gameplay. Mechanics are first pass, tuning is very rough, animations are placeholder and there are plenty of bugs. This is a glimpse behind the scenes at what the game looks like right now. It will get less prone to griefing, more filled-out and more polished as we continue to work on it. Okay, enough disclaimers. Here’s the video:



Explanation of what you’re seeing for those who want it:

The first clip is just showing me running into another player while wandering around the world. You probably noticed the near-constant “item picked up” sound effect: there’s a bug right now where whenever a player picks up an item, it propagates that sound to all other players. You probably also noticed that it’s horrible to listen to--most of us ended up playing with the sound off. The next clip shows a player dying and subsequently turning into a ghost. The third clip shows me as a ghost, haunting various items and another player (who correctly observes that I’m a ghost!). When I haunt something, it gives me back some life force (the blue stomach meter). As long as I have some life force, my humanity (the blue heart meter) won’t drain further, which is good because my humanity level is my maximum health if and when I get revived. If I manage to get my life force meter full (through haunting enough objects, which I was actually able to do in a dense forest--though the forest also got burned down, or by haunting a resurrection object or being given a reviver by a living player), then I will get resurrected. After that, you see a bit more chat in action. Then I show how resurrection works with the existing resurrection methods, specifically the Meat Effigy. We end with a shot showing even more players not starving together. This may have been the driest paragraph I’ve ever written.



On Playtesting Don't Starve Together

Don’t Starve Together is a somewhat difficult game to playtest. It takes a relatively long time to have a “meaningful” play session of Don’t Starve (and, by extension, Don’t Starve Together). Longer than 30 minutes, certainly. Typically, the main goal of a playtest this early in development is to suss out whether or not the core gameplay loop (or some aspect thereof) is compelling, which tends to be discoverable in less time than it takes to play a session of the full game. Because we’ve already proved that Don’t Starve works, a playtest of Don’t Starve Together is slightly different. We still are interested in whether or not the new mechanics we’re adding function well and fit with the rest of the game. But we’re also looking at social dynamics and how players interact, which isn’t necessarily something we’re making explicit mechanics for. Instead, we want to tune our mechanics to encourage certain types of play, but not strongly direct players towards that behavior. When there are 15 or so people joining a game with the knowledge that they had only a short amount of time to play, it seemed like everyone looked for the shorted path to having fun (a perfectly human thing to do!). Rather than starting to establish a base that would be quickly abandoned, most of us went straight to attacking each other and then, once dead, started griefing other players as a ghost. This was made worse by the tuning on my first pass of haunt-reactions being off the mark for certain reactions (notably: lighting stuff on fire). It’s important for us to bear in mind that, even though we don’t want to encourage people to play like this, we do want to allow for it. It shouldn’t be appealing or easy, but it should be possible. Functionally, this means making the bad effects of haunting into something you’re surprised by, not something you’re expecting. When talking about the playtest afterwards, the team decided that while the large playtest was useful in a lot of ways (it showed us that we can probably support larger numbers of players, among other things), it wasn’t really a good approximation of how we anticipate most people will want to play Don’t Starve Together. In light of that, we’re planning on doing a smaller and longer playtest sometime in the coming weeks. It will probably be four of us deliberately trying to cooperate and survive together for about 2 hours.



The Playtest: What Happened Back There?

Even though we’re going to do a more focused playtest sometime soon, it’s still worth examining what happened during this larger one. So, what did happen? I already ran through what’s shown in the video, so this section will be more about how the playtest was conducted and sharing some anecdotes. We started the playtest by having everyone join a dedicated server hosted in San Francisco, including Joe, who joined from his home (which is about 30 miles from the office). The movement speed of the ghosts was double the normal player speed during the playtest, which caused the server to start lagging a bit after a number of players had died. We decided to move to a new server on the local office network, which sorted out the lag issues.


When players joined the server, they would all spawn at the same location (the pillar with eternal fire you saw in the video), which made it really easy to start fighting, but difficult to start gathering resources: the immediate area got strip-mined pretty quickly. Some people immediately sensed the danger and jumped through a nearby wormhole to strike out and get established by themselves, with the intention of returning to the fray once they’d gotten some good gear. Others were perfectly happy to engage in the melee, turning to griefing as a ghost once they’d died (during the playtest, ghosts couldn’t actually perish, so they could do this forever--I’m working on that now). Fires raged and forests were lost. A few people, once they’d been turned into a ghost, actually started to cooperate with living players, following them around at night and providing just enough light to survive with. After one ghost in particular had lit the sapling or tuft of grass I was headed for on fire multiple times, I gave up on gathering generic resources and started to look for the ingredients for the reviver item, which would let me bring that pesky ghost back to life and make it harder for him to continue to grief me. While I didn’t succeed (I was cooking on a campfire because I was about to starve and the ghost haunted the fire, extinguishing it, which made me get stuck in that action), the player with the ghost-light friend was actually able to craft the reviver and bring his friend back to life.


Overall, there was definitely more griefing going on than we’d like, but we saw a pretty good range of behaviors. Let’s also not underplay how important this is: it functioned.



The Playtest: Takeaways

The two biggest things getting in the way of normal play were, above and beyond, the ghost-griefing and the strip-mining of the spawn area. Fortunately, both of these things should be pretty easy to address.


For the griefing, the first thing I’ll try is to take a second pass on the tuning of the percentage chance for the various possible reactions. The main problem here was that for many of the objects that could ignite as a reaction to being haunted, the chance for that to happen was 50%. When the thing goes up in flames half the time, it’s not surprising: you’re expecting it. Counting on it, even. You get into a pattern of haunting stuff with the hope of burning it down, which is extremely prone to griefing applications. That’s not the pattern we want. When you haunt something, you should expect a relatively neutral or even slightly beneficial result (i.e. an animation plays with no effect or you help chop down a tree). But there’s always the looming threat of something disastrous happening (i.e. starting a forest fire or spawning a Treeguard). If doing a second pass on the haunt reaction chances doesn’t cut it, I intend to try making haunting stuff actually drain your life force (which might also mean reworking the conditions needed for resurrection). I don’t anticipate that being necessary, though.


To deal with the strip-mining of the spawn area, I’m currently putting in a system that will put a handful of spawn points on the map at world gen, and then rotate through them as players join the server. Depending on the size of the server, you still may join and get placed in an area where another player also started, but chances are good that they’ll have moved on and the basic resources will have refreshed by the time you arrive. Things like flint and rocks are the exception here, since they don’t currently regrow like twigs, grass and berries. We’re completely aware that this will be an issue for games that have been running for a while, but we’re focused on getting the basics right before we address issues that crop up over longer periods of time. The scattered spawning (vs. single spawn point) will likely be a server option: that way, if you’re playing in a small group and know that you want to play cooperatively, you can start together, rather than spending time trying to find each other while also surviving. Personally, I’m more excited about the scattered spawn--for those of us who were able to escape the fray and get out into the world during the playtest, the experience of randomly happening upon another player always made for an interesting moment of uncertainty, similar to what you see in Day Z. These encounters work really well with the current locational chat mechanic (i.e. your typed message just appearing over your character’s head), which was also popular.


Even with the frustrations about the griefing that happened, reports were quite positive about the ghost mode. People enjoyed the level of impact you can have on the world as a ghost, as well as the way in which you can interact with stuff (and the density of possible targets). Being a ghost wasn’t boring once people understood how it worked, so we seem to be on the right track there. We do need to do a better job of instructing how the life force/humanity/haunting relationship works, but that too should be pretty easily solved. It was also really cool to see people very quickly discovering slightly non-obvious ways that the ghost can help out living players while they try to revive get the materials to resurrect the ghost player, like acting as a permanent and spooky torch.


On the technical side of things, there was some pretty bad lag when playing on the server in San Francisco, we’ve yet to do much optimizing on that front and we’ve reduced the ghost speed, so it’s not much of a concern. Lag aside, we were able to play with 15 people! That’s really exciting for us. We’re still not sure what the largest server size that we can support is--honestly, it may end up being smaller than 15--but this is a promising start. Additionally, we played on a dedicated server. There’s still some work to be done to make that work without any hacks, but that’s also a big win.


Lastly, I’m going to mention one more time: it works. A lot of our progress isn’t super tangible, but we’re steadily making it.



Next Steps


I’ll be addressing the issues I discussed above. This week, an artist joined the team and she’s off to the races making unique ghosts for the different characters, so we can all stop flying around as Abigail. The rest of the team is continuing to make the under-the-hood stuff work more robustly and completely.


Yesterday, the whole team sat down and had a discussion about PAX, which starts on August 29. We’ve got plenty of work to do before we will have a build that we can demo at the show. We made a list of the tasks that need to get finished in order to get to that build. We looked at the list and all agreed that it seemed doable. So now we’re all working with that goal in mind and prioritizing our work accordingly. We’re not sure we’ll get there, but it’s always useful to have something concrete to work towards. I’ll let everyone know for certain once we know ourselves if we will in fact be showing Don’t Starve Together at PAX. In the event that we do manage to get the build ready for PAX, please understand that that doesn’t mean it’s ready to go out into the wild: the alpha will come after PAX, regardless.


One last thing: the smaller playtest that I mentioned earlier in the post is planned for Thursday, July 17. I’ll try to get some footage and impressions from that playtest as well. But, fair warning, outside of that, there probably won’t be much in the way of these fireside chats until after PAX: we’ve got lots to do, I’ll be in Israel for a couple weeks on vacation, and these posts do take a fair amount of time, so I’d rather spend my time making my best effort to get us to a PAX build. That’s all for now, thanks for reading! As always, keep the discussion coming!

Link to comment
Share on other sites

Looks good. I fear the ranged weapons though, because the ice wand + gunpowder trick is easy and effective.


In my PvP experience (Nether),  the chat system was the best part. Though running into another player was terrible because I was always a target. Maybe DST can change my opinion about the ridiculous player "interaction" that those other games have. Getting killed for little reason and starting ALL OVER just kills my enjoyment in the game. 

Link to comment
Share on other sites

Just wondering, as people were talking about it in the above posts, will each character have their own Ghost? Or do you have to guess who it is and hope you're reviving the right person (in the sense of say, 2v2) I know it's more of a cosmetic thing but I am curious. 


Edit: Ignore me I have no reading comprehension 6/7 days a week.

Link to comment
Share on other sites

Yeah they are but I don't know why he wouldn't be able to go to Israel if he was Canadian. 

You cannot contain Canadians, we'll apologize our way right out the front door.

Klei is Canadian!? My brothers!


This certainly explains Woody though.

Link to comment
Share on other sites


This topic is now archived and is closed to further replies.

Please be aware that the content of this thread may be outdated and no longer applicable.

  • Create New...