Deception: Murder in Hong Kong review

Deception: Murder in Hong Kong (2014)

Game Details
NameDeception: Murder in Hong Kong (2014)
Accessibility ReportMeeple Like Us
ComplexityMedium Light [1.58]
BGG Rank252 [7.45]
Player Count (recommended)4-12 (5-12)
Designer(s)Tobey Ho
Buy it!Amazon Link

TL;DR: It's excellent! You should make every possible effort to try it!

I sometimes try to describe Deception: Murder in Hong Kong to people. I say ‘It’s basically CSI: Dixit’ . Sometimes I say ‘Columbo if it were set in the world of One Night Ultimate Werewolf’. The Dixit comparison in particular is quite odd though because it simultaneously captures perfectly the way I think of the game and it also comes across as bananas because it seems completely left field. The truth is that really Deception is only related to Dixit via a transitive properly of similarity. Deception plays like a streamlined version of Mysterium, which in turn plays like an overly bureaucratic version of Dixit. Deception is my favourite of the three. In other words, Deception: Murder in Hong Kong plays like the final deadly form of the Dixit World of Warcraft raid boss.

Deception box

It’s also by far the thematically weirdest of the games. That’s not at all easy since the competition includes a whimsical game of dream interpretation expressed through the scoring metaphor of rabbits. The framing of Deception seems utterly baffling – like a workplace disciplinary hearing expressed purely through the medium of angry subtweets. It plays beautifully, but it feels a little bit like it was a different game at one point that just got a lick of vaguely Film Noir paint before it was frantically stuffed into the box.

It’s super weird.

The setup is familiar to anyone that’s played a social deduction game – each player gets dealt out a secret role and it’s the job of the table to ferret out the miscreants. In this case, a murderer. Everyone else except for one player will be an investigator. The last will be a forensic scientist. For larger player counts you may also include an accomplice and a witness, but I’ve never actually felt the need to include them except to meet the top-end of player requirements. Their roles are predictable – they add information content into the game while also giving everyone a chance to obfuscate and lie about the roles they have. Essentially they act as an accelerant on the conversation.

Murderer clues

Each player gets four clue and four murder weapon cards dealt in front of them. As investigators these represent I assume whatever things they found at the crime scene and gathered up in the hope that they were relevant to the murder. You as an investigator come across as an overly eager rookie desperately trying to please an avuncular but ultimately disinterested superior who just wants to get home in time for Strictly Come Dancing.

‘Look Sarge, I found an ashtray at the scene of the crime!’

‘That’s nice, lad. Put it with the newspaper, the remote control and the junk mail you found earlier.’

‘I also found some lint. It’s oblong, approximately 1 centimetre in diameter. It’s blue! IT’S BLUE SARGE! Could that be important?’

‘Put it with the rest’

When the game begins, everyone closes their eyes except for the forensic scientist. The scientist then instructs the murderer to open their eyes and indicate a clue and a murder weapon from the set in front of them. For the purposes of our exemplar case our murder points to a sniper card and a computer.

If the witness and accomplice are present they’re also given access to some subset of this information. Everyone closes their eyes once that’s been determined, and then everyone opens their eyes once more.

At this point the forensic scientist deals out six tiles that represent a series of questions that nobody is asking about a crime nobody committed. The only way the forensic scientist can communicate with the investigators is through the placing of bullet tokens on specific clues in these categories. Their hope is that they can direct them towards the murderer along with the correct clue and correct weapon.

Evidence tiles

Seriously, how is this not a disciplinary matter? It’s like the scientist went in a massive huff at the start of the day because nobody noticed his new tan and then decided to sulk moodily in the corner instead of doing the job for which he was paid.

‘Can we get a ballistics report from you, Dr. Gunfinder?’

‘The victim was pretty fat.’

‘Uh… okay, but what about the ballistics?’

‘The murder happened in a park, or at least a park is the best way I can describe it from these six arbitrary options’

‘We just want to know about the calibre of the bullet – what the hell is wrong with you?’

‘If you don’t know, I’m not going to tell you. A bystander noticed a funny smell.’

It’s, as I may have mentioned, super weird. I hope the scientist is a member of a decent union with a good militant rep because I don’t see them surviving in the job for long otherwise with this kind of attitude.

This all means that the minute the game begins it descends into a kind of hilarious farce. If you’re the scientist you have to select between a range of options, none of which are good, to give clues that are at best incredibly ambiguous and non-specific given the range of cards in front of each player.

Second set of clues

It’s not enough for an investigator to identify the murderer, they also need enough evidence to convict. They need to get everything right from a set of clues that are likely to end up being so misleading that all the scientist can do is giggle to themselves. Those giggles will intensify as the investigators get more and more strident with every single trickle of useless information you send their way. When a bullet is placed, it’s stuck there – you don’t get to change your mind about its placement. That means if you send the table down a dead-end all you can do is put your head in your hands and hope that somehow the murderer is so strident in their allegations that they draw suspicious as a mere consequence of their advocacy.

Third set of clues

It’s not like your choices are great, after all – nothing is so specific that it will nail any given individual and even it was enough to precisely identify a combination of clue and weapon it’s the murderer that decides what the murder was. If they look around and see a lot of stabbing, slicing weapons then they’ll want to pick something sharp and pointy of their own. If there are a lot of office equipment cards spread around they’ll want their clue to be a photocopier as opposed to ‘a menu from a fancy restaurant’. They want to pick the combination of things that yields the greatest uncertainty from any clue you might give.

Cause of death - loss of blood

That means as the scientist you’re left trying to find the path of minimal damage through your clues as opposed to the path of maximum revelation. The murderer is a sniper, so what’s the most appropriate cause of death? If you pick ‘accident’ you’ll send people looking everywhere but at the sniper. Sniper implies a very specific professionality of murder rather than ‘Whoops, sorry!’

It’s not illness, it’s not poisoning, it’s not suffocation – it’s severe injury or loss of blood. But severe injury is very ambiguous and loss of blood wasn’t the cause of death but it might be closest? So, probably already laughing, you pick the one that seems most likely to drive people towards the proper culprit.

And then the table erupts in discussion and your job becomes an act of rapid triage to try and cut off the wrong lines of investigation and keep people fixed on true. Except, of course, you don’t have the necessary clues to do that and so everything you end up doing will likely make things worse.

Fourth set of clues

‘Well, it’s not me’, says the murderer. ‘Nothing I have would lead to a loss of blood as a cause of death. Pauline has a machete though, and also bites and tears! Bites and tears sounds like the most likely one of the table’

Oh no, you think. That’s both compelling and completely wrong and you don’t want the murderer to cement an early reputation for reliability of analysis. You put down a second clue. The duration of the crime was ‘instantaneous’

‘That rules out my bites and tears’, says Pauline. ‘Would probably take a while to hack someone up with a machete too’. Everyone sitting within hearing distance of your play group might be looking over a little warily at this point. ‘Snipers are pretty instantaneous though’

‘Death by sniper isn’t by blood loss’

‘But what else would it be from that card?’

‘Severe injury, obviously. Say, Jasmine has a pistol – that’s more likely to result in death by blood loss than a sniper rifle.’

Damn that’s also probably right now you think of it. Severe injury is non-specific but it doesn’t actually rule out the sniper in the same way you maybe did with loss of blood. Can’t change it though – the investigation is picking up speed and heading for the side of the mountain and you need to start steering it towards safer territory. You put down the next clue – ‘noticed by bystander: nothing’

Noticed by bystander: Nothing

Surely that’s the right thing? I mean, if it was a pistol someone would notice a sudden sound. Sniper rifles make a hell of a noise but they’re usually a long way away.

‘I’m stumped’, says the murderer. Pauline nods. Jasmine points over to Roz. ‘Throat slit? You could do that pretty covertly and all you’d hear would be a gurgling noise if you were close enough’

God damn them all.

Location: Building site

A building site! Come on, everyone – it’s the only one there that would be large enough for a sniper to get to work. I mean, you probably could shoot a sniper rifle at someone in a bookstore but it’s a bit excessive, right? Come on, come on, come on…

‘It’s Roz!’, exclaims Jasmine. ‘She used the acid, which she’d get on a building site, and she has the ID card she’d need to get in and out. It’s big and secluded and there are lots of places that she could murder someone without being seen or heard. I’m making an accusation!’

The tables looks expectantly at you. You grimace. ‘Hand in your gun and badge’, you say.

‘I don’t have a gun…’

‘Well, just your badge then’


As a minor aside – I’d recommend people buy a copy of Cash and Guns. Not for the game itself, but rather for the foam guns you get in the box. They are a prop that make several games a lot more viscerally satisfying.

Okay, final clue. Trace at the scene was… oh god, none of it is any good. None of it is any use. Nothing is going right. ‘Blood stain’

‘Well, duh’, says pretty much everyone. ‘The cause of death was loss of blood’

It’s okay though! Once all the clues have been placed the forensic scientist can replace one of the clue tiles with another drawn randomly from the stack. Not the location or the cause, but any of the others. You can get rid of that useless trace clue at the scene and replace it.

Hint on corpse: Head

The ‘hint on the corpse’ was ‘head’ because SNIPER, GOD DAMN IT EVERYONE.

‘Sniper!’, says Jasmine. She can’t make another guess but she can still contribute to the investigation. She becomes the obsessive, ranting disgraced cop that is still trying to get The Brass to listen to reason before she goes all Lethal Weapon 2 on the asses of the Mob.

‘Sniper!’, says Roz.

‘Pistol!’, says the murderer.

Thank god, you think. They finally got it. You did well, you did so well. You… you forgot to provide any evidence as to the computer clue and now they’re going dramatically off script. They’re fools – they’re working on the assumption you knew what you were doing and that the clues you placed must actually relate to the computer somewhere. If they were smarter they wouldn’t trust your nonsense. They’re the real villains in this piece.

‘Must be the telephone’, says Roz.

‘The circuit’, says Jasmine.

‘The computer’, says the murderer knowing that it’ll be instantly discounted because the table has now turned against him.

And all you can do, as you realise the murderer is almost certainly going to get away with it, is hang your head in shame and await the recriminations that come at the end of the game when you explain your, for lack of a better word, ‘reasoning’.

‘Fine!’, you say as everyone grumbles at the escaping murderer. ‘It’s not as easy as you think. Let’s see how well you do!’. You pass the cursed forensic scientist card to another player, or simply hope it doesn’t come your way again in a random draw.

The new scientist then proceeds to do very well because that’s just how cruel life can be at times.

I said at the start of this review that out of Dixit, Mysterium and Deception it’s this one I like the most. For me, it has exactly the right balance of all the key moving parts to bring out genuinely interesting game states. More importantly, fail states are hilarious and they’re hilarious well in advance of the actual failure. The curve of the game is always towards being enjoyable and the style of the puzzle it presents you keeps everything tractable without actually making it any easier.

In Dixit and Mysterium people can undertake whatever flights of association and interpretation they like and that’s tremendously freeing. It’s also something that doesn’t actually work especially well for some groups because playing cards skillfully is fundamentally an act of empathy. If you don’t know a group well the scope of your analysis is so open it can paradoxically be quite claustrophobic. If you know a group very well the act of analysis can be perfunctory – a natural common vocabulary of play will develop that results in the game becoming predictable. These games thrive in the median space between these two extremes but in aiming for that you will inevitably, through repetition, skew the experience of the group into sub-optimal territory. The games don’t get richer with repeated plays – they become frayed at the seams and increasingly threadbare.

Deception handles that well with the model of consent it adopts in the setup of the mystery. You’re not given cards and forced to react. Drawing the murderer card in Deception isn’t like being the spy in Spyfall or the werewolf in ONUW. It’s not like being a traitor in the Resistance. It’s more creative than that – you’re not given a role, you’re given a puzzle. ‘Which of my cards do I select to make the job of the scientist harder and my job as a murderer easier?’. In framing the job that way it creates an explicitly adversarial setup in the interpretation that keeps it fresh and interesting. Familiarity doesn’t converge on predictability. Instead it converges upon an energetic metagame that ensures the game retains a solid core of electricity. In Dixit and Mysterium repeated plays with the same group will result in certain things becoming shared tropes – memes that are constructed from the collective conscious activity of the table. In Deception you never have to worry about that.

Different evidence tiles

the structure of the game ensures that there are systems in place for actively preventing convergence on the same familiar lines of play. No matter what clues the scientist is giving out there is someone actively pointing out the alternatives. In the book World War Z, which I recommend everyone reads because it is awesome, mention is made of the ‘10th man doctrine’. If nine intelligence operatives agree with an analysis it is the moral responsibility of the tenth to stave off groupthink by taking up and advocating for the opposite position. Here, you have a similar kind of setup – no matter the consensus formed by the table there will always be at least one person challenging the narrative. Either it’s the correctly accused murderer trying to direct the light of inquiry on to someone else or it’s a falsely accused investigator trying to reflect it onto the person they think is the murderer. More often, there are multiple narratives being constructed and all have a certain credibility because of the non-specific evidence the scientist can give. Dixit and Mysterium rely on a peculiarly isolated form of creative interpretation. Deception gives you the fire that comes from friction.

The act of interpretation in Deception is also a good deal more deliberate and meaningful. If I want you to reflect on a colour in Mysterium and you reflect instead on the presence of a symbol that’s a disconnect between our frames of reference. Neither of us are wrong and while that’s very collegiate it’s also not particularly interesting. Deception permits players to build a coherence and a logic around their deductions. ‘It can’t be X because Y’ rather than ‘Well… maybe he wants us to think about the colour of the water?’. It’s more logical than empathic and while that may be a drawback for some people it does ensure that the discussion around the clues can be genuinely productive rather than merely hopeful.

Deception: Murder in Hong Kong is a fabulous little game that draws heavily from common gaming conventions and still manages to produce something distinctively unique as a result. We recommend it wholeheartedly.


One Night Ultimate Columbo absolutely should exist and I will literally throw money at that Kickstarter if it happens. You know Bezier, if you’re taking requests…