All posts by Eyejinx

Worlds Enough And Time

No Man’s Sky is an ambitious project.  It is one of the ultimate combinatorial extrapolations: 15 quintillion worlds, which is only true to the extent that a reddish world and a slightly more orange one are different worlds, or that a slight difference in topology meaningfully distinguishes one billion of those worlds from the other billion that are identical in everything except that one difference in the static noise variable.  Nonetheless, in its broadest strokes, No Man’s Sky does offer an unparalleled opportunity for exploratory and gathering gameplay.

One of the earliest lessons in my professional game design career was realizing that it is much more difficult to capture what works about a game than what doesn’t.  The errors are obvious; they jump out at the observer.  Bugs disrupt play, poorly implemented gameplay generates frustration, and as pattern-matching experts, we can always pick out something that isn’t quite right.  But, a good designer learns from what works and not just what falls short.  So, for the moment, I’m going to leave aside all of the obvious problems (crashes, repetition, opacity of systems) and focus on what works.

For one thing, NMS hits the basics: camera, controls, and character.  I’ve been playing the PS4 version, so no comment on the PC implementation, but on the console, the acceleration on the joysticks is tuned well, the FoV is what I expect from an FPS, there’s always a clear indication of what I’m pointing at, and I can adjust the key settings (inverted-Y, for example) to match what I prefer.  Not showing the player character is actually a valuable omission; given the variety of terrain and dealing with real-time morphing geometry, the animation problems of a third-person character would be a constant distraction.  As opposed to an RPG, the loot is not physical, so I have no need to see it on my character.  Similarly, omitting gravity variation from the combinatorial mix was an excellent call.  I’m sure the team must have experimented with it (given they are using Havok physics so get this practically for free), but navigating has enough challenges without dealing with gravity variations, and the animations of the ambient creatures would have been difficult (if not impossible) to get right with a wide range of gravities.

Exploring the world on foot is one of the primary modes of gameplay, and again, lots of good implementations here.  Sprinting and using the jetpack, for example, layer complexity onto movement without requiring their use or imposing penalties for failure to master them.  A beginning player can move with just the joystick; a more advanced player can use the joystick, the x button, and the sprint toggle to maximize the character’s speed across the landscape.  It’s also a solution to a long-term design problem: how to make walking/running as interesting in gameplay terms as driving.  Connected to this is the local scan function.  Rather than having the player wander arbitrarily around the landscape, the scan function polarizes the map, pointing out resources of interest.  Particularly in the sometimes extreme conditions of the harsher worlds, maximizing how much you can gather in a minimal amount of time outside your spaceship combines the navigation mechanics gameplay with a traveling salesman problem.  Yes, it’s procedural gameplay, but it works surprisingly well.  Any planet you land on will have some combination of resources distributed across the landscape, so playing hunter-gatherer is always an option.

Another win for NMS is the UI.  It’s a testament to the clarity of gameplay that most of the time, the UI fades off of the screen (with the exception of the compass, which is key to always have handy).  When you are in combat, your health and shields show up; when you are in hazardous conditions, an exposure timer shows up; when you are low on a key resource, the game alerts you, but most of the time, the UI is invisible – literally or figuratively – which is exactly what you want as a designer.  Managing the inventory screens is a little more clunky, which is inevitable given the combination of slot-based space management and a console controller; it’s probably seamless on PC.  But, again, the team at Hello Games did a great job of layering in complexity.  For a starting player, there are fewer slots to manage, and the slots occupy more of the screen space, making them easier to interact with.  As players progress and upgrade equipment, more slots become available, which would make it harder to navigate except that the players have developed sophistication in working with the system by that point.  Again, for most players, this elegance is invisible, but for game designers, take note.

On a personal level, I think NMS does a great job leveraging combinatorial systems in its spaceship designs.  Clearly, there are fixed body and wing types and some rules about how they can mix and match, but within those constraints, the system generates a wide variety of different-looking ships.  Sure, some are just re-palettized versions of the same geometry, but there are a lot of asymmetrical and intriguing combinations that come from the base inspirations – whether natural or artificial.  Associating dominant types with species and systems helps give the individual space stations more sense of place, as do the three different races and their variations in body and building morphology.  Similarly, the ambient alien lifeforms have a wide-ranging variety of shapes, sizes, and modes of locomotion.  The patterns become clear after a while, but the variation can create moments of discovery and wonder, essential to the sense of exploration.

Perhaps one of the most fundamental design aspects of the game is its commitment to opt-in gameplay.  As a player, you can choose at any given moment to play the ship-gathering game, the ship-fighting game, the planet-exploring game, the resource-gathering game, the trading game, or the galaxy-traversal game.  All of these aspects are readily available to the player, but only one of them is forced (ship combat cannot be avoided in some contexts).  Otherwise, it is up to the player to decide what her/his goal is, often how to accomplish that goal.  Aliens can be interacted with or avoided.  Quests do not need to be completed; they don’t block progress or hang around in a list forever.  The downside, of course, is a lack of direction, but that’s a player preference.  You can’t please everyone with the same game, and anyone looking for a directed, cinematic experience is going to be sorely disappointed.  But, the game is committed to its opt-in approach and creates opportunities rather than gates for the player.

There are, of course, all sorts of problems with the game, from performance to systems design to player communication.  With a game this ambitious, there are always going to be gaps; some are even generative – inspiring players and designers to come up with their own ideas about how to add or change something.  But, it’s practically impossible to smooth over every rough patch, eliminate every punishing set of conditions, or anticipate every combination with a game this ambitious.  A practiced professional could quibble with the game-makers’ decisions until the cows come home, but NMS fundamentally succeeds in what it sets out to do.  As a player, you can wander from planet to planet, gathering and exploring to your heart’s content.

For those of you who may still be learning the game, here is my advice:

  • Space = power.  Expand your exo-suit, multi-tool, and ship as quickly as you can.
    • Once you get your AtlasPassV1, there is an exo-suit upgrade terminal in every space station.
    • You can also look for drop-ships on any inhabited planet to get exo-suit upgrades.
    • Similarly, multi-tool upgrade stations can be found in many inhabited buildings.
      • Multi-tool upgrades are also the rewards of various alien interactions.
    • Finding a crashed ship is the cheapest way to upgrade your ship.
      • Search for transmissions at waypoints.  Transmissions can lead to observatories which can lead to ships.
        • Crashed ships will be at most +2 slots relative to your current ship.
        • You will have to repair major systems when swapping, but you can get away with just thrusters and engine if you want to work your way through a series of ships on the same planet.
    • Limit which ship and exo-suit upgrades you are using early on so that you can maximize cargo space.
      • Remember that one slot in the ship = two slots in the suit in terms of capacity.
      • You can always teleport resources from your suit to your ship; you can only teleport in the other direction when close by.
      • On the other hand, the only downside to upgrading your multi-tool is the resources it takes.
        • You may get multi-tool upgrades faster if you have filled most of the slots on your current one.  This is speculative, not proven.
  • It’s the economy, stupid.  You are going to burn through lots of units, so always be acquiring something of value.
    • On the ground, oxides and isotopes are the key to survival, but silicates and neutrals are the key to making money.
      • Also, don’t overlook planets with objects on them (like vortex cubes – 25K units/slot)
    • Look at the value of items you find.  If they are worth less than an equivalent stack of what you can gather, ditch them.
    • You can mine asteroids for rare neutrals – they’re the more pill-shaped ones.
      • You must have space in your ship inventory to collect while flying; the resources won’t go into your suit inventory (although you can manually transfer to free up space).
    • It’s good to build up a war-chest of 1-2 MM units early on.
      • There will be times when it is faster to buy components/resources from traders/terminals than to gather them.
      • Maxing your suit takes 6 MM units, and the costs continually climb, so having a million units on you helps avoid pinches.
      • A 30+ slot ship costs 7-8 MM units.  I’ve seen ships cost as much as 30 MM.
  • Combat is a resource competition. If you’re killing the enemy faster than it’s killing you, you will win.  Otherwise, run.
    • You can always replenish your ship’s shield by dumping oxides into it.
      • Always keep a stack on hand (iron is the easiest to gather, but titanium is more valuable)
      • Don’t wait until your shield is almost down – time doesn’t stop in inventory mode.
        • Depending on your upgrades, you may want to recharge at anywhere between 20% and 50%.
    • Make sure you’re attacking the right ships
      • Sentinel ships on top of pirates can quickly become overwhelming
    • On foot, an upgraded mining beam is just as powerful as a boltcaster.
    • In general, if you have enough shield, it is better to kill the sentinels and collect their resources than to run.
    • Stepping inside any building will cancel the alert level of sentinels.
      • You can practice combat or farm sentinels by camping an outpost.
      • Similarly, practice ship combat near a space station if you want an easy out.
  • If you’re following the Atlas path, do not sell any Atlas stones.
    • You will get 10 and need 10 when you reach the final step.
      • If you sell any before that point, you will need to replace them, and it costs 20X to buy them vs. what they sell for.

 

What Does $1MM Buy in Game Development?

A million dollars seems like a lot of money, but what does it really buy?  To explain the answer, I’m going to use a widely reviled concept: the man-month.  I should also clarify upfront that I’m talking largely about commercial game development; indies do more with less through a variety of approaches, but for most consumers, their points of reference are mainstream commercial games.  So, if you’ve ever wondered what it costs to build that kind of a game, here are some outlines.

The man-month concept gets a lot of grief; it’s called mythical, sexist, outdated; game developers throw shade at it all the time.  From a project management perspective, this is thoroughly justified.  Everyone knows that nine women can’t incubate a baby in one month, and using man-months to try and figure out the scope of an in-progress or near-future feature is like using a chisel to brush your teeth.  But, at the project planning stage, man-months can be a useful estimating tool, as long as you understand that they cannot account for critical path issues, do not reference actual men, and have low precision.

At the 10,000 foot view, when trying to establish the starting parameters for things like scope, budget, and schedule, the man-month is a valuable concept.  At a base level, a man-month is the rough amount of work that a team member can get done in a month.  Far more interesting is looking at the man-month cost of a project, though.  Let’s say, for example, that a project is going to take 5 people 10 months.  You can call that 50 man-months, and if you have an average man-month cost for your team (or studio), you can take that product, add some pad for additional overhead, and have a rough cost for the project.

What is an average man-month cost?  This is all over the map, depending on what kind of environment and team you are looking at.  The biggest part of this is always salary.  Let’s say you’re paying someone $84K a year; on a monthly basis, you’re paying $7K in salary.  However, salary generally only accounts for about 70% of an employee’s cost.  In addition to salary, you have benefits (insurance, vacation, etc.), rent (if you have a physical office), hardware (if you provide it), software licenses, etc.  So, that person making $84K a year is actually costing you about $10K per man-month.  That’s pretty good in today’s market for an A-level studio.  Indies are obviously cheaper; top-notch outfits go higher ($15-20K per man-month).

$84K a year is good money, above the established mark for where income makes a noticeable difference in quality of life.  Most people would be happy to make that kind of money.  The kicker is that on larger teams/projects, that’s the average salary.  So, let’s say you need a Tech Director for your project, and let’s say you manage to score one at the relatively bargain-basement price of $140K/year.  That additional $56K has to come from somewhere; if you take it all at once, someone else needs to be making $28K/year (under the poverty line), or you can spread that out across multiple people, but even spread across 4 other people, each of them has to take a $14K/year hit.  That’s for one hire.  For everyone you hire at a salary above average, you have to pull that money out somewhere else.  If programmers are pulling above $84K/year (and they generally are), your designers and artists are going to have to come in lower.  For every senior person, you’re going to need some juniors.

So, getting back to the question at hand, if you’re spending about $10K a man-month, and you have $1MM dollars to work with, you can afford about 100 man-months.  You can slice that various ways, 10 people for 10 months or 5 people for 20, but let’s set a timeline of a year; at 12 months, you can afford to hire 8 people (yourself included if you’re part of the project budget).  Let’s say off-hand that you need to have a lead for each of the major disciplines – tech, design, art, and production.  That leaves 4 additional hires.  Let’s say you want a full-time concept artist, modeler, and animator; that leaves you with one slot, probably a programmer.  Keep in mind that all of those people are going to need to be junior because your leads are going to need to be senior, and the man-month cost has to balance out.  Oh, and did you account for QA?  What about audio? Level design? Finance/accounting? Facilities management?

There are lots of ways to skin this cat; some projects don’t need a lot of art and/or can get more volume at less cost by outsourcing.  On a lot of teams this small, people wear multiple hats, because they have to.  Distributed teams are becoming more common and can avoid a lot of overhead costs, like facilities.  But, the bottom line is that for $1MM, you can get a small, 8-person development team that’s about half experienced people for about a year.  For contrast, most AAA PC and Console products take 50+ people to make; market leaders are almost all over 100 people, and MMO’s and other massive franchises can get to 200+ and multiple sites of development.  Even AAA mobile teams these days are commonly 20+.  Time to market is generally at least a year, even on small mobile projects (250+ man-months),  AAA is generally 2-3 years (1200+ man-months), and MMO’s run 5-8 years (4000+ man-months).  Remember, $1MM only buys you 100 man-months.

Or in other words, “not much”.