Valve has just announced a new 'Jungle Inferno' update for Team Fortress 2, with the highlight being a brand new map. Dubbed 'Mercenary Park', it's a "new jungle-themed disease-ridden three-control point map" created by Valve. 

The video below doesn't really show much of the map, it being a new animated short providing some narrative colour to the proceedings. But the map isn't all there is: there are a handful of community-created maps as well, all jungle-themed, and for modes including Attack / Defend and King of the Hill.

There's a bunch of new taunts as well: the Yeti punch is what the name implies, and there are a tonne others too, all listed neatly over on this update page. 

This is all just the "Day 1" content – expect more stuff to roll out or be announced over the coming days.

Valve's next TF2 update is going to be a biggie, overhauling some of the multiplayer shooter's most iconic weapons and fixing items that have been unbalanced for years. When Valve announced the changes in June it didn't say when the update was due. After a blog post from the developer yesterday we still don't have a precise date, but we do have a much better idea.

"We're putting the finishing touches on a mammoth new update, and it'll be shipping in the very near future, we promise. How near? Well, very. Imminent. Not this week imminent, but you know. Really soon."

Hmm. The language suggests that it could arrive later this month. I really hope so, because it's the first TF2 update for a while that I'm actually excited about. There's tweaks across the board but some classes, such as the Scout or Spy, are getting a lot of attention (the Scout's triple jump-enabling Atomizer bat is now much less viable, for example), while others like the Demoman are being left mostly alone. 

You can read more about the changes in Valve's "sneak peak".

Every player will have a different view of the planned changes, but personally I'm looking forward to a nerf to the Spy's Dead Ringer, a cloaking device that players use to feign death that is really frustrating to play against. Ammo kits and dispensers will no longer fill the device's meter, so Spies won't be able to use it as often.

How do you think the changes will impact the game?

The Orange Box launched ten years ago. It was undoubtedly the greatest bundle of games ever, with the simultaneous launch of Portal, Team Fortress 2 and Half-Life 2: Episode Two, alongside the existing Half-Life 2 and Half-Life 2: Episode One. The former three were instantly significant in the landscape of PC gaming: Portal was an influential puzzle game that many cited as the surprise highlight of the set, while Team Fortress 2 arrived as a fully-formed multiplayer phenomenon that would constantly evolve across the next decade. Episode Two, of course, was the last time we experienced a new chapter of arguably the greatest singleplayer FPS series of all time. 

It was a massive moment: imagine that many amazing games dropping at once now, from the same developer. It just wouldn't happen. Here, Valve's Robin Walker reflects on the factors that led to The Orange Box's release, and offers some behind-the-scenes insights on both Portal and Team Fortress 2.

PC Gamer: What did the release of The Orange Box mean for Valve at that time, and what does it represent as part of Steam's history?

Robin Walker: The Orange Box was a huge step for us internally because it was the first time we’d ever managed to complete more than a single product at a time. In some ways, the Orange Box was a company level 'hack' where we made three separate products that all consider themselves the same product for shipping purposes, which meant that people could rationally prioritize their work across all three of them. If you were on Portal, and everything was going well, but TF2 was struggling, it made sense for you to jump over and help TF2 out because all three games needed to ship together.

The Orange Box was also a great product to really highlight why the retail channel was reducing game developer’s options. We found with Episode One that retail really didn’t understand or like a premium quality $20 title—they stood to make less money per box, and they had a limited amount of shelf space in their stores. The Orange Box avoided this by combining multiple quality products into a single box that was worth that full amount, but in doing so it created other problems. Retail had never seen a new, high quality box containing more than one title. Historically, a box that contained multiple titles was a bundle of old or low quality titles.

So in terms of Steam’s history, to us the Orange Box represents the era in which distribution channels placed a huge amount of friction on what kinds of games were made, how big they should be, and how much they were sold for. These weren’t things that retailers should be blamed for, they were simply the side effects of operating in physical space. It’s great to be able to look around and see such an enormously wide spectrum of games being made today, many of which wouldn’t have had much of a chance to find their audience in that physical distribution world. 

Were you surprised by the response to Portal, in that a lot of people considered it to be the highlight of The Orange Box at the time?

We didn’t really know what to hope for with Portal. We’d put it in front of enough play testers to be confident that players would have fun with it, but Portal didn’t fit any existing model of a successful game for us to know how it was going to really turn out. There wasn’t much of a history of first person puzzle games, let alone ones that combined a new gameplay mechanic with comedy. The Orange Box really solved Portal’s biggest challenge, which was to explain itself to players. By putting it in the Orange Box, we didn’t have to do the heavy lifting of explaining to people why they should buy this thing that was unlike anything they’d played before—instead, we could lure them in with Episode Two & TF2, and surprise them with the game they had the least expectations for.

Portal became incredibly influential to the indie games scene—its length, storytelling and environmental design are felt in a lot of today's games. Can you recall that process of the Narbacular Drop team joining Valve, and the key decisions that eventually made that game what it is?

By the time we saw Narbacular Drop at the Digipen student day, we’d already hired multiple groups of inexperienced developers who had built interesting things. When we hire those kinds of teams, we’re fundamentally more interested in the people than the thing they’ve built, and in our discussions with them, the Portal team seemed like a group of people with a huge amount of potential. We paired them up with some experienced developers at Valve, and let the team loose.

In any game's development, there are too many decisions to count, and many of them will ruin the game if made incorrectly. One decision that ended up being very important was the one behind GladOS. We had been working on Portal for about a year, and at that point we had 14 levels of the game in a state where they were being regularly playtested. There was no GladOS, the player just moved from puzzle to puzzle without any sense of progression or reward beyond the increasing complexity of the puzzles. The playtest response we kept seeing could be summed up as "This is really fun! When does the game start?". This was both great and terrifying. Players were having fun, but they seemed to consider everything they played as just training leading up to something else. Considering the entire game was really just a process of learning about the core gameplay mechanic, this scared us a lot, making us worry that we’d have to create a whole other section of the game afterwards.

But first, we asked ourselves what it was that was causing players to consider everything as training. After much discussion, we settled on the idea that it was the lack of threat or pressure. Nothing in the game pushed back on the player. There was no real failure, no cost to mistakes, nothing overall to fear, no larger goal to strive for, and hence no real reason to advance. We talked about various solutions, and in the end decided that introducing an antagonist made the most sense. The antagonist could start as a narrative tool for introduction & reward, and over time become the thing that pushed back on the player, eventually giving them the core goal of the game—"I want to learn all this because I need to be able to defeat X". We had little in the way of art production on the team, so it being a character that largely spoke to you via voice over was a straightforward production solution.

In the end, there are many important decisions after this that were critical to GladOS working as well as she did, such as her entire personality. But her genesis begins with a straightforward process of us trying to solve the core gameplay problem in Portal. Even today, it’s always fascinating to us that players seem to start Portal talking about the gameplay, but after they’re done, all they talk about is GladOS.

You've kept updating and transforming Team Fortress 2 over the years, and few competitive games have that kind of lifespan. What's been the philosophy behind that? How have you kept reinventing the game while still making it recognisably TF2?

The philosophy is pretty simple—listen to your players, pay attention to what they're doing, ship your work, and iterate as much as possible. But TF2's a strange thing. In some ways, it seems so different to how it launched in 2007, but at the same time, it still feels utterly familiar. There are still Snipers on the battlements in 2Fort having a fine old time paying no attention to what's going on with their flag in the basement. There's a much wider set of potential threats to deal with than they faced back in 2007, but they now have many more choices in exactly how they want to face them. And no matter what they decide, they can ensure they look different to all the other Snipers in the game.

So TF2's core gameplay seems to be fairly resilient in the face of all the horrible things we've done to it, and I think that's largely due to how we've approached our role in the process. We've always felt that our job was to support players in whatever they're trying to do. As a result, it's the players who've decided how TF2 should be played throughout the last decade. We've added all kinds of elements to the game, from both our and the community's minds, and the players have been the ones to digest and choose the way those elements ended up incorporated into the whole, even if it meant outright rejection in some cases.

You provided audio commentary for The Orange Box at the time, which was a really nice opportunity to let players get granular with the various games' creative processes, having previously tested it in Lost Coast. Can you recall the process of doing that? What was it like to examine your work through that lens as a developer?

We approached commentary as a tool to explain our craft. In our experiences listening to commentaries of other creative works, it was the nuts & bolts of how they actually did the work that interested us the most. Throughout our years of developing games, we constantly found that problems we thought were going to be straightforward to solve turned out to be nasty, thorny issues involving complex tradeoffs between design and technology. Often, that complexity was hidden entirely by the solution. So we thought it might be interesting to players if we could lift the rock and show them everything that’s going on underneath all that apparent simplicity. We’re game developers, so hopefully players will forgive us for thinking that game development is a fun thing to talk about.

Also, that commentary and accompanying analysis was all written before the product launched, which means we didn’t have the chance to examine our work through the context of how it was received, let alone how it would fit into the gaming landscape 10 years later. Would Portal be something people would like? Or would it be some weird puzzle game Valve made that no-one wanted any more of? Without that perspective, we found it hard to talk about anything other than what we were confident in—what we did, and why we did it.

This article was originally published in October 2017. We're repromoting it alongside re-reviews of some of the PC's biggest living games, looking at how they've changed over time.

Not long ago, every first-person shooter had to have a multiplayer mode. Part of multiplayer's job in the late '00s and early '10s was to extend the lifespan of your purchase—to keep Xbox and PlayStation owners' copies in their homes and off of GameStop's second-hand shelves after they finished a six or ten-hour campaign.  

It was a dark time for FPS players. We got underwhelming and under-supported multiplayer for games like F.E.A.R, Crysis, Singularity, Prey, Homefront, Far Cry 3, Medal of Honor, Call of Juarez, Duke Nukem Forever, Rage, and Syndicate. In the worst cases, the multiplayer modes of these games contorted the themes and designs into competitive forms. Spec-Ops: The Line lead designer Cory Davis described the multiplayer component of his game years later as "a cancerous growth," saying that "the game mechanics were raped to make it happen, and it was a waste of money."

One surprise to come out of this era was BioShock 2, which fully embraced the awkwardness of turning the intricate story and setting of Rapture into a deathmatch. You played as one of 10 unique characters living in Rapture at the outset of the civil war. Each had their own backstory, melee weapon art, and loads of custom, situation-specific dialogue that you'd hear during matches (Mlle Blanche de Glace, a French actress, might say "So strange on this side of the camera," when using the Research Camera to earn a damage buff against an enemy). BioShock 2 characters' appearance was tied to their progression—as they leveled up, they'd look more and more like a mutated Splicer, a reflection of their abuse of ADAM. The multiplayer lobby itself was a Rapture apartment, and your character would receive new, unique messages on their answering machine. 

The whole thing was framed as a prequel for the first BioShock, grounded in an idea that's only taken hold over the past few years: yes, you can and should put story into a multiplayer FPS.

Talking it out

Credit for killing the trend of tacked-on multiplayer should go to Team Fortress 2. In the 10 years since The Orange Box, Valve's spy-fi FPS has made plenty of contributions to the genre, to Steam, and to the PC. But most significantly, it helped teach the industry, and games like BioShock 2, how to tell stories in competitive multiplayer games.

We take it for granted now, but in 2007, TF2's addition of dozens—eventually hundreds—of situationally-triggered character voice lines was innovative. Without watching a cutscene or reading a dull menu blurb, players would be hit over the head with the idea that the Scout loves baseball ("I'm battin' a thousand!"; "Yo, I oughta' be on a baseball card!"), or that the Soldier is an over-the-top nationalist ("You are now a conscientious objector to being dead, hippie!" after dominating another, apparently lesser Soldier; or "Your white flag does not stop American bullets," to the Spy). 

The Spy has his own set of lines that mock the Scout after a kill. I like "May I borrow your earpiece? [mimicking Scout] 'This is Scout! Rainbows make me cry! Over," but the best one has to be "Well, off to visit your mother!" a reference to the canon relationship between Spy and Scout's mom (this literally haunts the Scout's dreams) and a special reward for sticking a knife into the fastest character in the game. We'd see a similar system in Left 4 Dead just a year later.

The voice lines are more than gags: they're economical hits of characterization, taken passively in the breaths between TF2's combat, that lessen the repetition of playing 2fort for the two-thousandth time. One by one, they form a tapestry of connections between TF2's nine classes, reinforcing the hide-and-seek game that Spies and Pyros play, or the symbiotic relationship between the Heavy and Medic.

Some of this is owed to how integrated Valve's writers were into the development of TF2 during this part of the studio's history. "[W]e don’t have a strict wall between design and writing," Valve writer Erik Wolpaw said in 2011. "We’re all together in the same room, designers and writers." You can see the fruits of this integration of writers in TF2's biggest updates, which made new patches not simply 'content,' but richly-themed events led by their fiction, rather than their features.

The massive WAR! update in 2009 solidified the template we'd see for years to come. Even by today's standards, the patch was an avalanche of lore, weapons, new systems, and excitement that made the seven days it lasted a special experience for players. The highlights:

• Kicked off a weeklong contest between Soldier and Demoman players. Via the TF2 wiki: "A kill tracker appeared in the HUD of anybody playing as a Demoman or Soldier, with the counter increasing each time they (as a Soldier) killed an enemy Demoman or they (as a Demoman) killed an enemy Soldier." Whichever side won would earn a new weapon for their class.
• Launched an official Propaganda Contest, calling for fan art (that in turn helped market TF2 to the wider gaming community) focused around the war between Demomen and Soldiers. The winner of the Propaganda Contest, J.Axer, received a one-of-a-kind cosmetic item, a Towering Pillar of Hats with a ribbon pinned to it.
• Six new items, for either the Soldier or the Demoman, were made available.
• Valve published a 21-page comic that formed a backstory for the update: the Demoman and Soldier had become friends, which in the eyes of TF2's in-game announcer, the Administrator, had to be put to an end.
• Published a series of web pages that colorfully housed all the information around this update, which also included easter eggs like the Shakespearicles portrait.
• Oh, and crafting was added.

Framing the addition of a new item as a time-sensitive dramatic event in the TF2 timeline that players themselves had to participate in in order to resolve was a brilliant act of marketing, and we see this concept replicated in competitive games like Overwatch, Destiny 2, CS:GO, and Dota 2. "A secret, exclusive seventh weapon has also been developed by the TF2 team," wrote the TF2 Blog. "In other words, either the Soldier or the Demo (but NOT both) is walking out of this update with one more weapon than the other. If you want your class to have it, you're going to have to fight for it." 

The Soldier and the Demo were natural rivals that, up until that point, hadn't had their yin-and-yangness formally expressed. They were both splash-damage classes with different styles: firing parabolically, or firing in a straight line. They could both use their weapons to propel themselves forward. An ordinary sticky bomb kill on cp_well, scored that week, had new meaning, adding to the overall body count.

All of this fun helped to soften the introduction of a new crafting system, laying the groundwork for TF2's item economy. Valve would use this technique again when it introduced paid items and loot crates a year later not as a storefront, but in the form of a fictional character entity: Mann Co. Although it wasn't without controversy and uproar, again, narrative framing and ample humor softened the landing microtransactions made on TF2, a much riskier and more novel concept in 2010 than it is today. You weren't buying items from Valve, you were buying them from a shirtless Australian.

Made to last

It's true that plenty of Team Fortress 2's storytelling has been done outside of its .exe, in comics, short videos, blogs, and marketing imagery. But in widening where its story was being told, Team Fortress 2 broke an old way of thinking, that a singleplayer campaign was the beginning and end of a game's fiction. It's part of the reason we've stopped seeing multiplayer shoehorned into shooters like Prey, Wolfenstein, and Deus Ex.

That approach has produced some of the most well-rounded and memorable multiplayer characters ever, ones that rival protagonists and villains from conventional games. When Blizzard announced Overwatch in 2014, game director Jeff Kaplan said “If people want to compare Overwatch to Team Fortress 2, we would take that as the world’s greatest compliment," said Kaplan. “We love that game; it’s probably one of my favourite games of all time. Those guys are geniuses," he said.

Not many FPSes are being played enthusiastically by a million-and-a-half people 10 years after release. But Quake, Unreal Tournament, Counter-Strike, Rainbow Six, and Battlefield don't have 15-minute short films about fighting giant loaves of mutant bread.

To celebrate 10 years of the Orange Box, we're publishing the original PC Gamer reviews from our archives. All three games were reviewed by Tom Francis in issue 180 of the magazine. The reviews capture our excitement at the release of a bundle containing three 90+ rated games. October 2007 was a great month for PC games.

Half-Life 2: Episode Two review

Last night I had a really weird dream. I dreamt that instead of ending where it ended, Half-Life 2 just carried on after the explosion at the top of the Citadel. Vortigaunts rescued me, then Alyx hugged me, and I ended up catching a train out of town into a countryside full of tiny Striders. Then a Combine Advisor made me do my uni finals all over again, only this time I hadn’t revised and the papers were all in Chinese. And I was naked. 

It’s surreal that Valve are still churning out more Half-Life 2, three years on. As beautifully crafted as Episode One was, it did tread on a lot of its parent’s toes. Episode Two certainly doesn’t do that. It turbos away from them at 90 miles an hour in a customised Dodge Charger, with Alyx riding shotgun. 

I won’t spoil any details, but Ep2 is what happens after you and Alyx break free of City 17 once and for all. The setting for most of your previous adventures is nothing more than a smouldering scar on Episode Two’s skyline, and the Citadel looks like a longfinished game of girder-Jenga. Because of that, and because you spend a lot of time driving a car that could have swerved straight off the set of a postapocalyptic Dukes of Hazzard, Ep2 feels wild, dangerous and cool. 

Your time—a little under five hours—is diced into refreshingly different sections. Valve still do pacing better than anyone. They break fights with puzzles, driving with combat, solitude with friendly faces and claustrophobic tunnelrunning with epic, sweeping vistas of naturalistic landscape.

These make Ep2 feel huge. It only took me an hour longer than Episode One, but every inch of it is gorgeous uncharted territory, and there are more inches than the running time suggests. Spending a third of the episode in a supercharged two-seater means covering a lot of ground, of course, but it’s more than that. There’s an openness to a lot of Episode Two’s chilly forested landscapes that’s new to Half-Life.

It’s all the more inviting because Episode Two is the most sumptuous chapter of the Half-Life saga, and by a country mile. It’s as if Valve’s tech and art teams are trying to outdo each other: the Source engine has had a striking technical overhaul that renders textures, materials and curves uncannily well, and the artists clearly relish having a fresh palette to work with. 

Towering conifers bristle gently in the breeze, casting soft shadows across winding mountain paths. Each toothy vortigaunt’s big peering eye glints glassily, a perfect ruby sunk into finely wrinkled brown skin. Even the shotgun is newly beautiful, gleaming ominously in the sun with a convincingly weighty gunmetal sheen. We get to see the pine-covered rocky land of this nameless nation, and it conforms to no established gameenvironment stereotype. It resembles only the real world—some proud, cold country I feel sure I’ve been to—and it has that authentic real-world grubbiness that only Valve have figured out how to recreate. 

The combat explodes across this soothing canvas with a brilliantly messy splat. Something clever involving particle physics has allowed Valve to make thick black blood, lurid yellow goo and something a lot like vomit spray repulsively from your victims with every cracking impact. The new poisonous Worker Antlions burst like bioluminescent bombs; injured Hunters drool a sticky slurry of their own innards from where their mouth should be; and when the vortigaunts fight... Jesus God. The trailers released last year showed nothing of this—some consolation for those of us who spoiled big chunks of the game for ourselves by watching them.

The three-legged Hunter creatures are the highlight of the fighting: velociraptors to the Strider’s T-rex. They’re the perfect size for Gordon-killing: compact enough to chase you indoors but hefty enough to take the shotgun blast that awaits them there. More importantly, they’re bright enough to do so when you least expect it. Valve have trained them to deduce where you’re heading and get there first by a different route, and the effect is alarming. 

They’re another departure from what we’ve previously been shown of the game: they used to fire a tiresomely familiar pulserifle burst, now they fling a torrent of bulbous azure darts on swirling trails. These thud into whatever they hit, quiver pregnantly for a moment, then detonate in a flurry of fizzling plasma pops. Cover is no longer enough: wherever you hide, in a few seconds you’re going to have to throw yourself from the room before it explodes. 

Once they’ve flushed you out with their flechettes, the Hunters use Episode Two’s wide open spaces to take a scampering runup and smack you into the stratosphere. The seemingly generic name is exactly apt: they take the initiative in combat, and nowhere is entirely safe. The Hunters continually provoke and surprise you in ways that Half-Life hasn’t since the first game’s marines. 

The sense of threat is a prevailing and escalating theme As you can see, the pistol has been beefed up a little. Once, just once, itd be nice to have a windscreen. This is what Vorts do to Antlions. of Episode Two, and it extends to the plot. You and your friends are trapped, maimed and violated in ways that are distressing on a really visceral level, and it’s properly gruesome to watch. The blood-soaked tone gives the story a force that makes it the darkest and most exhilarating chapter yet. It’s Half-Life’s Empire Strikes Back—and it even has a less snowy analogue to the Battle of Hoth.

That fight needs to be mentioned—but mustn’t be described—because it’s the first truly satisfying climax to a Half- Life game. Half-Life’s Xen was disastrous, shunting energy balls at a bigger energy ball in HL2 was uninspired, and taking down a single Strider with the RPG at the end of Ep1 was almost comically banal. But at last Valve have crafted a finale worthy of the adventure that precedes it, and the result is the largest, most open-ended and complex battle of the series. It’s also one of my favourite Half-Life moments so far, and that’s saying a lot.

Episode Two is, needless to say, so polished that it hurts to look directly at it in sunlight—so my only criticisms are pretty feeble. The first is that it’s slightly too easy, right to the end. I wouldn’t mind if Hard mode only increased the damage you took. But it also reduces the damage you deal, and that renders almost all the game’s weapons meek and unsatisfying. Speaking of which, we still haven’t had a single new gun since the end of Half-Life 2. That game never went nine hours without introducing several new weapons, so where are our shiny new deathsticks? The armament is the only part of the Half-Life formula that’s starting to go stale. 

Episode One seems dismally small and boring by comparison, however much I loved it at the time. But the one edge it does retain over the second is Alyx: she’s not quite as charming here. There are a few really wonderful character moments with her, and one superb performance from actress Merle Dandridge, but nothing quite as heart-meltingly cute as Episode One’s Zombine joke. To be fair, that’s only because the grim plot mostly keeps her in her less convincing :o and :( emotional states. 

Still, I didn’t think it was possible for a mere episode to surprise, excite and energise me as much as it has. The simultaneous global unlocking of the next chapter of Half-Life is becoming one of the most geekily enjoyable events on the gaming calendar, even with the mortifying suspense of relying on Steam under heavy load. Not just the playing, but IMing inanities to friends while it unlocks with an agonising slowness, and retiring exhaustedly to the nearest forum afterwards to exchange breathless superlatives. “Oh, and the bit where—and then she—and the Advisors!”

Half-Life 2’s critics groan about its linearity, and its fans groan about how long it is between episodes. But the series shrugs off both complaints when you start seeing it for what it is: a series of playable movies. Better than anyone, Valve make cinema that you’re a part of—and they do it at about the same rate Warner Bros churn out new Harry Potter films. 

In Half-Life’s day, it was a compliment to say a game felt like a Hollywood movie. In the intervening years Valve have become more professional and accomplished than any visual effects studio working in Tinseltown, and now it would be an insult to liken one of their beautifully crafted works to the messy dross the American film industry churns out. 

I know some gamers love a sandbox—I’m one of them—but it always baffles me when that love seems to preclude the enjoyment of anything else. Do these people storm out of cinemas when they realise their popcorn munching isn’t controlling the actors? Are there really people who can’t enjoy a gorgeous, hurtling ride like this? That would be sad, because they’re getting seriously good. Bring on the third.

Verdict Fresh and thrillingly darkScore 93

Team Fortress 2 review

Last night I had a really weird dream. I dreamt that Valve were finally going to bring out Team Fortress 2, only they’d made it look like some crazy Pixar cartoon, it was budget-priced, and all the classes talked as you played. And I was naked. 

But here we are. This is the trouble with a dream job: you have to do it even when you’re asleep. I’m just going to review this ridiculous ‘Team Fortress 2’ fantasy until I finally wake up and discover that I am, after all, a chartered accountant. 

Let’s not dwell too much on the original mod for Quake and Half-Life—that was ten years ago, not everyone played it, and TF2 is very obviously aimed at new players as much as old. Worth mentioning, quickly, is that it’s got the same nine classes but fewer weapons for each, grenades have been removed entirely (thank God) and, well... look at it. Look what they did to it. 

The changes might sound like simplification, but like the art style it’s more about exaggeration. The Spy used to have a double-barrelled shotgun, for goodness’ sake. Taking stuff like that out hasn’t made it a simpler game, it’s made the choice to be a Spy a more meaningful one. Every class is so tightly focused on doing its thing that TF2 feels like nine different games fighting each other. That’s bewildering at first, but it’s a joy to watch characters this beautiful smash each other to pieces while you learn.

That Pixar comparison isn’t fair. TF2’s gurning murderers look better. Valve have remodelled their class-based multiplayer FPS after the work of turn-of-the-last-century illustrator JC Leyendecker. Google Image him and you’ll see the similarities in the angular, characterful silhouettes. They’re a world away from the lumpy sacks that were The Incredibles, and as it turns out, class-based multiplayer combat has long needed that distinctiveness. 

It sounds like a small thing, to be able to tell what class someone is as surely and as clearly as you can see them at all. To have an immediate sense of the heft and power of a Heavy, rather than an abstract notion of his hitpoints. But stuff like this has an intensifying effect on your moment-to-moment experience: you feel, see and comprehend the game world in Technicolor. It makes all the relationships instantly clear and the importance of your actions explicit. In short, it makes everything you do 300% cooler. 

That’s Team Fortress 2: multiplayer magnified. Cooperation means more, victory is sweeter, betrayal is more bitter, defeat more humiliating. But it’s what lies at the heart of multiplayer gaming that matters most, and that is, in the parlance of our times, the lols. 

The image of a Scout circle-strafing a Heavy quickly enough to smack him into a stupor with a tiny baseball bat is inherently funny. But it only really gets a belly-laugh when the Scout is a scampering stickboy in knee-high socks, and his victim a meat-headed brickheap of a man. Character is a catalyst for comedy, and until now multiplayer games just haven’t had it. They were already funny, but TF2 just brings it out beautifully, every round.

All that stuff—gloating, humiliation, snuff slapstick—is best with friends. But another of TF2’s charms is that you form relationships with the people you’re playing with so quickly. They might not be friendships exactly, but they add an edge of human interest to every interaction. I don’t know Gabe Newell very well, but after he’d followed me around as my personal Medic for a while, I felt like I did. The same goes for Robin Walker after he and I—as Engineers—constructed an elaborate ecosystem of killing machines that reaped dozens of enemy lives.

That’s a quirk of the way friendly classes tessellate, but TF2 is more interested in playing up your relationship with players on the enemy team. Each time you die the game freeze-frames on your killer after a short delay, and that delay is calculated very cynically to catch him in the middle of an offensive taunt animation. Worse, the game then invites you to save this lewd image of your murderer for posterity. And the game looks so damnably good that you’re usually compelled to do it. 

Valve know we like to mock the dead, dance on graves, hump corpses. So as well as making that mockery more crushing, they’ve also made a game of it: taunting now roots you to the spot, pulls you out into third-person view to watch yourself swagger, leaving you utterly helpless. You’ve actually got to make a strategic decision about whether you’ve got a few seconds to play air guitar on your victim’s carcass or not. I’ve seen chain-reactions of death where a Sniper waves to his unfortunate victim, is shot dead mid-mock by another, who then performs the same taunt—with the same fatal result. 

The taunts, and the lines uttered alongside them, are part of the persona Valve have given each class. If you’ve seen the Meet The Heavy or Soldier trailers, you’ve had a taste of this. (See the disc for the most recent.) But the idea that your character is a character, with his own personality, is only as relevant as you make it. If you leave the taunt and chat commands alone, you’ll only really hear yourself if you’re a Heavy: the big guy can’t resist cackling deliriously if you’re getting a lot of kills, and an extraordinary spree will usually be punctuated with a bellowing “SO... MUCH... BLOOD!”

The other classes’ involuntary comments are too quiet and infrequent to hear often, but in a quiet moment I did hear the liquor-chugging Demoman mutter that “On the plus side, I already don’t remember this.” If, like me, you develop a particular fondness for one character, you can hammer the chat shortcuts and taunts to mutter battlecries at every opportunity. It was the gasmasked Pyro I fell in love with, and if you’re wondering what his voice is like, the answer is a punchline in itself: muffled. 

His battle-cry is “Mmmph mm mumph umph!” and his call for a medic is “Mmphumph!” His dumpy teardrop physique, shrew-like tiny head, waddling gait and baggy, flame-retardant suit—they all evoke an endearingly downtrodden man. So I run everywhere garbling incomprehensible insults, rocking out on my fireaxe over my crispy fallen foes, and waving my bent petrol-pump flamethrower exultantly over my head after every match; win or lose. “Mmmph mm mumph umph!” 

Most maps kick off with the two teams separated by a metal mesh that lifts after a minute, giving Engineers time to build their defences and everyone else a chance to taunt each other. The result is two rows of people jeering, singing, laughing, braying, dancing and whooping at each other in a cacophony of clashing voices. It’s a long-needed outlet for our natural tendency to pre-game smack talk, and it makes the atmosphere of the calm before the storm electric. 

TF2 comes with six maps; three are new, three are remakes. The roster doesn’t feel slim once you play them. Hydro, a control-point map split into six zones, restructures itself between rounds to put teams into one of 16 different configurations. The others are mostly a linear series of control points—all except Gravelpit, which gives attackers a choice of two to assault, and 2Fort, which remains stubbornly Capture The Flag. Capture The Intelligence, sorry.

All are heartbreakingly gorgeous. The soft lines, gentle shading, warm palettes and wonky edges set off the gaudy action magnificently. It’s tempting to pussyfoot around with weasel words such as ‘among’, ‘could be’ and ‘in years’, but screw it. Team Fortress 2 is the most beautiful game ever made. I say that as a man who’s seen Crysis running on maximum settings, and I’m not kidding or exaggerating or on any more than the usual amount of drugs. Sorry about that, every other videogame artist in the world. This was not your lifetime. 

Granary and Well are a little straightforward flow-wise, but TF2 sometimes benefits from a simpler arena. A linear series of control points might not sound like a lot of fun, but the simplicity shifts the focus to tactics and clever use of classes. My team won a round when another player crept behind the enemy lines to camp their locked-off control point, capturing it the moment it came into play. This despite the fact that he—presumably inadvertently—announced his plan to the entire enemy team by using ‘say’ instead of ‘team say’. The post-game chat revealed that they didn’t take it well: 

EricS: We’re actually going to have to erect flood barriers from all the QQing going on over here. 

Finole: I need a raft. 

Robin of Death: Driller’s quit FPSs. 

Robin of Death: And the company. 

Gravelpit is a more interesting equation: the defending team has to guess which of three tall towers the attacking team is going to gun for first. Dustbowl, meanwhile, is sure to be a cult favourite all over again: it retains the deafeningly chaotic opening, the succession of increasingly bloody chokepoints and the desperate last stand. 2Fort is faithful to the classic original in all but appearance, and makes a particularly rich playground for the more tightly focused classes.

Hydro’s more like a set of maps than a single arena, and feels a little arbitrary for it. Valve have made it this way to keep it fresh, but I’m not sure multiplayer maps need to be fresh. CS’s de_dust is great partly because we all know it so well—so is 2Fort, for that matter. Hydro’s hypermagical rejiggling just extends the period for which you’re not really sure where you’re going. We won’t know for months whether it was worth this to keep mixing things up, and I’m happy to bear with Valve’s experiment for now—some parts of Hydro are superb. 

The initial confusion of Hydro does highlight a real shortcoming of TF2, though: no minimap. Only the most co-ordinated, voice-communicating hardcore clan has any useful notion of where their team-mates are. In a game where the nine different classes are so interdependent, it’s vital to know where your turrets are set up, whether there’s a Medic nearby, and if the Heavy cavalry is on its way. Valve’s logic in omitting it is that some players never use them. Fine for those guys, but there’s no substitute for the rest of us to knowing where our team is without having to ask everyone all the time. 

Mind you, a minimap would make the Spy’s life harder. He was always Team Fortress’s most inventive class, but his new incarnation is even more extraordinary. He can disguise himself impeccably as any class of enemy, and now he can also render himself temporarily invisible to slip into their base. There’s no friendly fire in TF2, but shooting all your team-mates to uncover Spies wastes too much time and ammo to be practical. 

As a Spy in disguise you still take damage from enemies, but you’re man enough not to show it—you don’t bleed. That gives rise to a hilarious mindgame: a good Spy will take a near-lethal shotgun blast to the face from a supposed friend without flinching, confront his attacker toe-to-toe as if to say “What?”, and continue his infiltration beyond suspicion.

The Spy’s disguise-o-meter, built into his cigarette case, will give him the name of an enemy who really is the class he’s pretending to be. That means that every now and then, you experience the alarming existential crisis of encountering someone with your own name. Realising they must be an enemy Spy, you declare to your team that “The Spy’s a Soldier!” Whereupon, of course, everyone empties their magazines into you. 

If you can stay away from your namesake and take the Spy hunter’s check-shots unflinchingly, the challenge becomes to act like an enemy. I like to dress up as a Heavy, because his reassuringly enormous size makes it hard for anyone to believe he could be a slinky Spy in disguise. It also means you get healed by enemy Medics—a peculiar sensation—and that can lead to an utterly bizarre psychological dance. 

The Spy, you see, needs to get behind his victim for a one-hitkill backstab. The Medic, meanwhile, should always stay behind a Heavy for protection while he heals. So the two of you run in circles trying to get behind each other, until the Medic realises—with an almost visible pang of horror—who you really are. He draws his bonesaw, you draw your butterfly knife, and the duel commences. It’s sublime. The knife-edge between the thrill of deception and the shame of discovery makes playing a Spy more tense and thrilling than any other multiplayer experience—even the original TF’s Spy.

The other class highlights are more obvious: shredding a dozen enemies as a Medic-boosted Heavy, bolting past a superior force as a Scout with the briefcase in 2Fort, and detonating enough pipe bombs as a Demoman to fill the room with blood—and the screen with kill reports. In fact, the only class that doesn’t excite is the Medic. His contribution is to heal the major players while they charge in, but he can’t do anything else while he heals so his whole life is just holding down fire. When he’s healed a thousand or so points, he can temporarily make himself and his mark invulnerable, at which point he has to... keep holding down fire. It’s so cruel that he doesn’t get to let rip after all that joyless service to his team. 

I have to admit that this, and the minimap problem, bothered me less and less the more I played. Team Fortress 2’s friendly look hides what’s still a dauntingly intricate game, and when you’re still learning the ropes, and the maps, its few flaws seem exasperating. But the measure of a multiplayer game is how much you want to go back to it. Right now I’m quivering a little, and last night I dreamt about it, so yeah. Team Fortress 2 is a bit special.

Verdict Rich, gorgeous and endlessly fun.Score 94

Portal review

Last night I had a really weird dream. I dreamt that the students who made Narbacular Drop, the space-bending indie game I played last year, had been hired by Valve to make a beautiful new version in the Half-Life universe. You played this wide-eyed girl with strange metal braces on her shins like Eli Vance’s prosthetic leg, and a droning robotic voice kept saying sinister things that I found incredibly funny. Also my Year Nine maths teacher was there. And I was naked. 

But you’re probably here to read about Portal, Valve’s first-person puzzle game about opening rifts in space to cross uncrossable obstacle courses. It’s designed around one simple but mindexpanding idea: you can shoot a hole in any wall, and then another one somewhere else, and if you walk into one you’ll come out of the other. Fire them side by side and you’ll walk straight back into the room you just left. Fire them on floor and ceiling and you’ll fall through the same room at terminal velocity forever. 

The Aperture Science Handheld Portal Device (let’s face it, the portal gun) is your only weapon, but its brazenly impossible uses are endlessly fascinating to perform. The internal logic is flawless, but you somehow never grasp it entirely however hungrily your brain gropes. It is, I’m just starting to realise, abstractly kinky. Portal, honey, how can your physics be wrong when they feel so right?

The game grips you by the wrist and leads you briskly past the befuddling basics of these rifts, straight to the good stuff. Within a few short levels you’re using orthogonal portals to translate your gravitational potential into lateral velocity and flinging yourself exhilaratingly over turrets and lethal slime. By nudging you gently through rooms that cleverly lead your eye to the correct—yet patently impossible—solution, it swiftly teaches you a dazzling roster of lunatic tricks. 

Portal is a magnificent puzzle game. The titillating wrongness of every solution and the wonky thinking required to get there make you feel like a space-folding genius, and yet you’ll almost never get stuck. Soon you’ve learnt so many weird ways of perverting the forces and spaces in any room that you can throw yourself through them, like a futuristic Prince of Persia with abilities more improbable and wondrous by far. 

The solutions eventually become more gymnastic—opening new portals mid-fling and plummeting back through those you’ve previously opened with pinpoint precision. But by then you’re ready, and performing deliciously counter-logical mental inversions at breakneck speed is something to be relished.

The atmosphere, meanwhile, grows thickly sinister. Your singsong robot guide GLaDOS (you’ll find out what it stands for) doesn’t seem unduly invested in keeping you alive. Soon her own delusions creep into her instructions to you. “The weighted companion cube,” she announces as you snatch up a box, “will not threaten to stab you and cannot, in fact, talk. If the weighted companion cube does talk, the Enrichment Centre urges you to disregard its advice.” But as her coldly voiced lines become more murderous and surreal, they also get funnier. The writing is effortlessly sharp throughout, and with its single inhuman character Portal taps a thick vein of black, absurdist humour that becomes the game’s propulsive force. You’ll play faster just to hear the next beautifully unhinged line. 

The game escalates magnificently. The puzzles change nature, requiring your to beat the system with the tricks it taught you rather than jumping through hoops. And at the same time, the humour reaches fever pitch—GLaDOS becomes so brilliantly deranged that at times it’s hard to control yourself for laughing. The final flourish—the most inspired credits sequence I’ve ever seen—reduced me to a convulsively cackling wreck, insensible and almost in tears. This is the funniest game I’ve played since Psychonauts.

Sadly, Portal is as short as it is sweet. It took me—admittedly a Narbacular Drop veteran, international super-agent and one of the greatest minds of our time—a little under two and a half hours. That’s long enough for the story it tells, and it tells it well, but it’s so damnably good that the craving sets in as soon as the satisfaction fades. 

Depending on gamer demand, Valve say they’ll opt next for a straight sequel, a closer tie-in to the Half-Life games or some form of multiplayer. I just want more GLaDOS. Her lilting, darkly comic words of lethally unhelpful advice deserve a place in the annals of scary robo-speak, right beside “I can’t do that, Dave” and “L-look at you, hacker.” 

“If you begin to feel lightheaded from thirst,” GLaDOS chirps, “feel free to pass out.”

Verdict Brain-melting offbeat genius.Score 92

Loot boxes are everywhere. They're in shooters, RPGs, card games, action games and MOBAs. They also take the form of packs, chests and crates. They're filled with voice lines, weapon skins, new pants or materials to get you more loot boxes. They're in free games and paid ones, singleplayer and multiplayer. They can be free to open and paid for with real money. You may feel an almost violent antipathy to the very idea of them, but you've probably also opened a fair few.

The appeal isn't hard to grasp. Opening a loot box is a rush: a moment of anticipation followed by release. That colourful animated flurry is often accompanied by disappointment, but is sometimes with the joy of getting exactly the item that you wanted. And then you feel the gambler's pull to open another, pushing you back into the game to grind or digging into your wallet to earn or buy your next one.

"It's that moment of excitement that anything's possible," Ben Thompson, art director on Hearthstone, tells me. "In that moment I could be getting the cards I've been looking for for ten or 20 packs. That anticipation has always been a key point in games in general; successful games build on anticipation and release, whether a set of effects or in gameplay."

Loot boxes' ubiquity might be fairly new, but they've been around rather longer than you might think. Economic sociologist Vili Lehdonvirta has suggested that they appeared in their modern form first in the Chinese free-to-play MMO ZT Online in around 2006 or 2007. A Chinese newspaper described how for a yuan you would buy a key: "When the key is applied to the chest, the screen will display a glittering chest opening. All kinds of materials and equipment spin inside the chest like the drums on a slot machine as the wheel of light spins." Yep, sounds like a loot box. 

But they've also been around far longer in the form of baseball cards and Magic: The Gathering packs, and, if you think about it, even in identifying magic items in D&D. In each case you experience the same notes of suspense and reveal, and also the way the reward is separated from the action you took to earn them. That's an important distinction. Loot boxes aren't quite the same as the shower of loot you get for killing an elite monster in Diablo. There's more of a build-up, and rather than being focused on moment-to-moment play, your view is being pulled out far wider, into the meta game, into the larger systems that give you reasons to keep swinging your sword.

The psychology of loot

Why do loot boxes provide such a dark compulsion? Psychologists call the principle by which they work on the human mind 'variable rate reinforcement.' "The player is basically working for reward by making a series of responses, but the rewards are delivered unpredictably," says Dr Luke Clark, director at the Center for Gambling Research at the University of British Columbia. "We know that the dopamine system, which is targeted by drugs of abuse, is also very interested in unpredictable rewards. Dopamine cells are most active when there is maximum uncertainty, and the dopamine system responds more to an uncertain reward than the same reward delivered on a predictable basis."

What's more, the effect of variable rate reinforcement is very persistent. Psychologist B.F. Skinner conducted trials during the early 1930s in which he conditioned animals to respond to certain stimuli in closed chambers that became known as Skinner Boxes, and showed that even when the rewards were removed, the subject would continue responding for sometimes hundreds of trials, trying to recreate the circumstances in which it got its reward before.

"Modern video games then amplify this idea by having many overlapping variable ratio schedules," says Clark. "You're trying to level up, advance your avatar, get rare add-ons, build up game currency, all at the same time. What this means is that there is a regular trickle of some kind of reinforcement." Whether you're watching your XP climb up to the next level in Overwatch, or you're collecting scraps in Battlefield 1 by breaking down skins, there's a constant sense of reward leading to reward.

The clever—or insidious—bit is how a loot box is wired into a game, and how it doles out its baubles, keeping a player on the knife-edge between feeling hungry and feeling rewarded. One such system is Battlefield 1’s Battlepacks. Standard Battlepacks are earned by playing multiplayer matches. They used to be randomly awarded, but they recently switched to an Overwatch-like progression bar system for more regular drops. Each one is a guaranteed weapon skin or one of a number of pieces of a unique weapon. So that would seem satisfying, if it wasn’t for the scrap system.

Here, you can turn your skins into scraps an in-game currency called Scraps, which will buy you more Battlepacks. And they’re the only way without spending real money that you can access Superior and Enhanced Battlepacks, two upper tiers which have rather better chances of dropping Distinguished or Legendary weapon skins. The result is a system which ekes out rewards and then asks you to question them and wonder: should you dispose of them in the interests of getting better stuff?

It’s a complex system with a lot to get your head around, and remember: Battlefield 1 is meant to primarily be an FPS, not a lottery game. In other games, loot systems sit more centrally, and few are more central as the card packs in Hearthstone. Since it’s a collectible card game, they’re perhaps so fundamental to the game that it's inaccurate to consider them loot boxes in the same vein as the controversial packs of skins and items added to recent big-budget games like Destiny 2 and Middle-earth: Shadow of War. Still, they're a great example of the loot box's principles. 

How Blizzard designs loot boxes

You can buy packs in Hearthstone with an in-game currency called gold. There are several ways to earn it, but the key methods are that every third game you win awards you with 10 gold, and for each daily quest you complete, such as winning games with a certain class, you'll get at least 40 gold. A card pack costs 100, so you can expect to earn at least one every couple of days. This system is subtly integrated into play; most quests gently encourage you to try classes and playstyles you're not used to, while also rewarding you for simply playing the way you like. Or you can just buy card packs with real money. Classic card packs cost $3 for two, $10 for seven, and the scale goes up to $70 for 60. Despite the pride some take in being free-to-play, most will spend money at some point, while those who don't get the reward of telling themselves they're saving money by playing. 

The five cards you get in each pack will be taken from across all the game's classes, at least one of which will be 'rare' quality. "We're just straightforward with it," says Thompson, but it has other benefits. "People are more inspired to try different and new things. So if I get a number of Shaman cards, maybe it's interesting for me to start to build a Shaman deck? Or I can craft them into cards I do want in the game. We allow player agency to dictate it, but we also avoid putting them in a position where they choose themselves out of experiences." 

The loot box's place in Overwatch is quite different since they contain cosmetic items—skins, emotes, voice lines and victory poses—rather than the very thing you play with. But you acquire them in a similar way: play with any character and you earn XP, with various bonuses granted, for example, by playing with friends and for good performance. Level up, which is possible every hour or so, and you earn a loot box.

"We aimed for players earning a box or two in a gaming session, so that you wouldn't walk away from a session empty-handed," principal designer Michael Heiberg tells me. "An earlier version of the game's progression system had per-hero experience levels, with rewards at various hero levels. In testing, though, we saw players picking heroes based on these hero level rewards instead of picking based on what the team needed, or even what they felt like playing. It was a bust, and we knew we needed to disassociate your hero picks from the rewards. Based on that, we shifted to a system with randomized rewards that you could earn by playing as any hero."

Overwatch's loot box is a masterpiece of audio-visual design. "It's all about building the anticipation. When the box is there you're excited at the possibilities of what could be inside," says senior game designer Jeremy Craig. Click the ‘Open loot box’ button and the box bursts open, sending four disks into the sky. Their rarity is indicated by coloured streaks to further build the suspense. "Seeing purple or gold you start to think about what specific legendary or epic you've unlocked. This all happens so fast, but it was those discrete steps that we felt maximized excitement and anticipation."

Hearthstone's opening animation is likewise engineered to trigger anticipation, and also to make the cards desirable objects and to imbue them with a sense of value. From the start it was important that they'd evoke real collectible cards. As Thompson says: "Ripping that foil pack and feeling it give, that moment of excitement that anything's possible."

Rather than hitting a button and watching, as you do when opening most loot boxes, from Battlefield 1 to Overwatch, you have to drag a pack over to what Blizzard calls the altar. There's a brief moment as blue magical power builds, and then, in the case of the classic packs, the cards suddenly burst out in a shower of glitter and gold. With Journey to Un'goro packs, they emerge in a crackle of lightning (which echoes its evolve mechanic), and a shattering of ice in the Knights of the Frozen Throne packs.

The challenge was to design a sequence that would feel special to those opening a single pack while not wearying those opening 50 in a row. "If you buy that many you don't want to spend half your day opening them, you want to get them open and start building decks and experience the real focus of the game," says Thompson. "As much ceremony as we want to put into the pack opening, we need to keep it concise." The sweet spot, it turns out, is about two seconds. 

As Overwatch does, Hearthstone indicates the rarity level of the cards you'll be getting before the cards are actually revealed. Mouse over their backs and you'll see a colored glow on rare, epic or legendaries. "We don't immediately flip them, we let player agency take a seat in the sense of controlling what order they flip them in, how they flip them, the time between each flip."

Loot makes you superstitious

That hint of control is quietly important to the design of Hearthstone's card packs. "What we found in talking to people is that superstition sets in," says Thompson. "What you'll find in psychology is that if the outcome is of high import, you know like, 'Gosh I hope I get a legendary in this,' and if player agency is unclear in terms of your ability to manifest any kind of change in the outcome and there's a little bit of randomness involved, superstition takes hold. That agency and sense of involvement and choice is super important in terms of the experience and the enjoyment of it." 

You've probably dabbled in something like it too, by performing some kind of personal rite before opening a loot box. Here's YouTuber Jordan 'Kootra' Mathewson mass-opening Team Fortress 2 crates his own way. This behaviour is actually common across many species: Skinner discovered in 1947 that even pigeons exhibit it. He observed that they’d practise little rituals in the hope that they’d cause food to appear, including turning around in their cages or nodding their heads, and yet the food was given to them at entirely regular intervals. The absence of any explanation of why the food appeared had conditioned them to believe their actions caused it. On a deep level, our own minds work the same way.

Overwatch and Hearthstone contrast with the common way loot boxes are presented. The Counter-Strike: Global Offensive model, in which the gun skins in the crate scroll by, slot machine-style, is a direct evolution of the old ZT Online design. Their distinct lack of visual pizazz is compensated for with the graphic way they show you what you could have won, and when the needle just misses the item you wanted, it's hard not to reach for another go, even though as far as CS:GO is concerned it's as black and white a result as rolling a die.

This design closely mirrors the near-misses in many forms of gambling, from horse racing to roulette. As psychologist Luke Clark has said, "A moderate frequency of near-misses encourages prolonged gambling, even in student volunteers who do not gamble on a regular basis. Problem gamblers often interpret near-misses as evidence that they are mastering the game and that a win is on the way."

In most countries, including the US and UK, loot boxes are not legally considered gambling because the winnings have no intrinsic value outside the game (in China, laws have actually forced developers like Blizzard and Valve to publish the drop rates of their loot boxes). But in being expensive to buy and based on the same psychological principles, we have to treat them with the same care.

Why do we love collecting stuff?

Loot boxes also plug into another facet of psychology: collection. In 1991, Dr Ruth Formanek in the Journal of Social Behavior and Personality suggested five reasons we feel the compulsion to collect, including 'extending the self' by obtaining knowledge or having sole control over one's collection, the social benefits of collecting leading to meeting like-minded others, creating a sense of continuity in the world, financial investment, and addiction or compulsion. Alternatively, Freud suggested that it's rooted in a deep desire to reclaim the poo you excreted as a baby. 

Whichever theory you go with, loot boxes are almost always filled with collectibles. Overwatch's boards of sprays and percentage counts for completion rates on characters remind you of what you've accrued, and Hearthstone is a collectible card game. Games as a whole highlight an interesting distinction between freeform and structured collection. Collecting, say, baseball caps is freeform collection because you can accrue them indefinitely. But games present a very structured form of collection, tapping into several powerful motivational principles. You're working towards a clear and achievable goal and you can see your progress towards it. During matches you get to show it off to others who are also immersed in collecting the same items, a chance to feel both kinship and bask in the status your collection confers. 

And there are systems of scarcity, driving value towards certain items. But managing them is a delicate art. "We use rarity levels primarily to control the frequency of getting our most exciting content," says Overwatch principal designer Heiberg. "We don't want players getting frustrated because they're earning none of the best rewards. We also don't want players getting bored because they earned all of the best rewards at once. Rarity levels give us some control over the pace of these rewards."

Both Overwatch and Hearthstone's designers are careful not to dictate value. "We learned that the value of our cosmetic content varies widely from player to player, and that no distribution of rarities was likely to really jive with everyone," Heiberg continues.

"Some players are super excited about that rare card and the legendary doesn't mean so much, and similarly you'll have someone trying to build an all-Murloc deck and they're going to be more excited about a common Murloc as opposed to the legendary of a class they're not after. We let those moments be fun at every level and not focusing on legendary cards being awesome and how you should get all of them, but rather let the player get excited about any aspect of the opening."

It's easy to feel uncomfortable with loot boxes. They have a powerful capacity to manipulate your behaviour and extract considerable amount of time and money from you with systems that aren't the core game you actually want to play. The bad ones use these tricks to make you value in-game items that you might not choose to in the cold light of day. They can pull you to do things to acquire them that you’ll regret in the long term. But the well-designed ones give you space to find your own value in the trinkets they dole out. That's an indicator that they respect you, and a sign that they recognise—correctly—that collection should be a reward in itself.

"Pack opening is an area that took a fair bit of time to develop because it's a moment players will spend a lot of time with," says Thompson. "More importantly, they'll spend money there and any time our players are investing time and money we want to give them a very fair and honest return. We want people to walk away feeling they got value from it, and that value can come from not just a return on that time or money but also fun. We say we make decisions in Hearthstone based on how much fun players are having, and pack opening is no less of that."