Shadowing Practice: How would Alan Turing have played the Traitors? - Learn English Speaking with Video

So you might have heard that I am part of this little TV programme called Celebrity Traitors, no big deal, but because I am a massive,

gigantic nerd, cut me open,

I bleed equations, I obviously wanted to approach this in the most scientific way ever.

So before I went into the castle I spent a frankly

ludicrous amount of time trying to come up with the most statistically perfectly perfect approach to catch me some traitors

and um and I think I found one I think I found something really good

so uh I thought you might like to see it

so you can know what is going on in my head

while I'm in the castle uh and here is what follows the mathematical way to try and catch a traitor.

Now one science you think that you might be tempted to try is uh the sort of behavioral clues of people lying.

This is what pop psychology tells you.

You can spot somebody's lying if they're sweaty or their pupils are dilated, or they look up and to the left as their mouth is moving.

Unfortunately, all of the evidence unequivocally says that you can't do it.

The TSA, who are the people who run airport security in America,

people whose literal job is to spot people trying to deceive them,

they once spent hundreds of millions of dollars trying to train

their officers to spot behavioral indicators of deception to spot lying essentially and then

and then there came a point where they were like should

we should we check whether any of this actually works

so they went through and they reviewed 400 different studies over six decades

and they found no valid scientific evidence whatsoever

that you could reliably detect lying

so basically they've ditched the whole thing the csa don't even bother doing

and if the professional security officers are not using

that tactic um well it's not going to work for me

so i'm going to need something else there is one other

interesting thing from a study in 2024 in korea where a group of scientists wanted to see

if you could catch traitors based on what they said

so not what their face was doing but their actual words

so they recorded people playing the game mafia

which is sort of like a dinner party version of traitors

and they ran the transcripts through AI

and asked it to predict who the traitors were

and it managed to do it much better than the humans

did it could predict the traitors 50 % more often than humans could

and crucially the AI got better the longer the game went

on unfortunately I think I think bringing in recording devices in an AI model into the car

so I think that's I think that's off the off the table I don't think I think Claudia will come

and strangle me with some leather gloves if I try and do anything like that.

So if psychology is out and if tracking people's words is out, I'm going to need something else.

I'm going to need another tactic to catch me some traitors.

So before I went into the castle, I had catch up with my good friend Matt Parker while we plotted the best strategies to use.

And the goal here is to determine who is a traitor.

Yes, and I think that the way that people have played this in the past is they're looking for definites.

You did this, therefore you must be a traitor.

Right.

And ultimately, I don't think that there are really hardly any signals that are absolutely deterministic.

I think you are looking at indications of direction.

Right.

So I wanted to run the maths on the cleanest, you know, verifiable signal.

You only want unambiguous, costly signals.

Absolutely.

None of this people just chatting.

None of this human nonsense.

Exactly, exactly like as we all aspire to you've approximated social

interactions with numbers to a few key binary decisions that's exactly what i've done okay before we get to

that though yes you can try and work out who is a traitor

but another way to approach this is to think about who is not going to be a traitor

or more accurately which combinations of people would not work together as traitors.

So if you imagine all of the possible universes of all of the possible combinations of three traitors

and then all I've got to do is eliminate those possibilities

one by one by one by one to hopefully lead me to the correct three people.

I should tell you I stole this method from Alan Turing.

This is exactly how he approached cracking the Enigma machine.

He started off with the list of all of the possibilities

and then eliminated everything that wouldn't work um to sort of get him to the final answer

and honestly if this is

if this is good enough for alan jerry i think i

think it's good enough for me um there's just just one

one tiny glaring problem there's 21 people there's three traitors the

number of possible universes is it's kind of huge it's 1330 however

what i can say from the off is

that i'm gonna ask to be a faithful

so i can um i can eliminate a lot of those universes where i'm included which then gives me 1140. so

when you start the game in front of you is a

huge space yes of 1140 possible realities you could be living in

and you're is is way too high to be useful

but luckily there are some ways to bring it down quite quickly i don't think

that claudia will pick three traders who are just all the

same person i don't think it's going to be three male

comedians i don't think she's going to pick three mathematicians all between 35

and 45 years old you know i don't think she is so

what i've done is i've gone through

and i've worked out called it a homogeneity flag i've worked out all the universes in

which all three traitors share similar traits

that are too similar for them to all be picked and

when you do that actually the number of universes come down really quickly

if you get rid of all of the universes with three comedians

or three actors or whatever it might be it drops down to 618.

now we're talking that's that's much more like it

but now here comes the big assumption because i don't think

that three traitors will want to hang out together i don't

think you'll see three traitors sitting next to each other at

breakfast i don't think you'll see three traitors sitting next to each other at the round table

and i don't think

that you will see three traitors sitting together in the cars there might be two traitors in a car sure

but i don't think you'll ever have three so

if there are four people in a car i get to delete four universes

because there's four choose three yeah those four

100

so they're all gone yeah those four universes are gone cross

101

them off there's five cards delete those universes um at breakfast

102

there's uh these you know four people in a breakfast group

103

delete those universes four at a time yeah four at a time

104

and then on the round table i can basically go around

105

and write right not those three those three those three those

106

three those three those three those three those it's gonna be like 20 -ish of them right exactly

107

so if this is what's happening on the start by day two

108

this number comes down a lot

109

thanks to those cars and those chance meetings but even more than

110

that it starts to come down

111

because people get kicked out of the game every day one person gets murdered one person gets banished

112

and all their universes leave with them all of their universes leave with them

113

so they start coming down coming down coming down by the very next day

114

if you started on 618 you go down to 442 live universes.

115

By day four you're down to 72 universes

116

and it's gonna be difficult you know to say

117

because sometimes people get into the same car over and over again

118

but I think

119

that by day five i think you'll be down to just a handful of potential universes

120

so now i've got this much smaller set of groups

121

which in theory i mean if this works means

122

that i should be getting quite close to who the traitors are by day five um

123

but remember i'm gonna i'm gonna try

124

and hit this from both ends okay as well as as well as noting all of the things

125

that traitors don't do and eliminating those possibilities i also want to be looking at the things

126

that traitors do do here's the thing

127

what you really care about is the chance

128

that somebody is a traitor i mean that's that's literally the whole game right there

129

and before anything happens before you get any information whatsoever you know

130

that the sort of baseline chance of this is three traitors out of 21 people in total

131

which is roughly 14 percent right at the beginning everyone's got

132

a 14 chance of being a traitor now what you want

133

to do as the game goes on as you get collectively

134

more evidence is you want to update your probability of somebody being a traitor now the temptation

135

when you're playing the game is whenever something happens you know someone cries at the round table

136

or somebody is caught out in a lie the temptation is to say oh well you did

137

that therefore you're a traitor the problem is the masters doesn't work

138

that way around because for every action

139

that a traitor might do there is a chance

140

that a faithful might do it as well can i give

141

you an example yes please of the messiness of faithfuls

142

so um i've called them burn votes so i've noticed this tendency of traitors to throw a vote, essentially.

143

When they're in the first three, four episodes, when they're feeling really anxious, they will just pick somebody, anybody, just pick.

144

Because they don't want to go with the majority.

145

They don't want to start their own thing.

146

They don't really know what to do.

147

They don't want to bring anyone with them.

148

They don't want to bring anyone with them, so they throw a burn vote.

149

So if you look in the first three

150

or four episodes at who has voted for somebody that nobody else has voted for, you will often find a traitor hiding there.

151

Or let me do it the other way around.

152

The traitors are always hiding in that group.

153

The problem is, is it's not a deterministic signal because the faithfuls are also just all over.

154

Also spraying boats everywhere.

155

Yeah, right.

156

Exactly.

157

So there is no way to statistically find anything that is definitely pointing towards a traitor.

158

Instead, you're looking for these traitor -ish signals. And given

159

that there are way more faithfuls than there are traitors in

160

the game i need to work out just how suspicious a particular behavior actually is

161

and the goal is to pin down as many actions as i can

162

that are more likely to come from a traitor than from

163

a faithful this by the way also absolutely ripped off from alan turing he knew

164

that that humans were operating this enigma machine and he knew

165

that humans are an absolute sloppy mess

166

so uh bletchley he kept a library of two million cards of everything

167

that happened up until that point which they then used to predict likely new messages

168

so what i'm going to need i'm going to need a

169

version of my own library here i'm going to need a

170

database of how humans have previously behaved during games of traitors

171

oh i've been doing some work now i've been doing things okay

172

so first of all what i did is i went

173

and uh collected all of the data from the previous british shows shows plural shows plural

174

so yeah i um i went back through every single previous

175

episode of uh the uk version of the traitors civilian

176

and celebrity and uh

177

and i took took a lot of notes insane amount of

178

notes frankly i fast forwarded through all the the wishy -washy chats

179

and the challenges and just logged every choice and action um

180

that counted as hard evidence

181

so you know we're talking votes we're talking seating patterns we're

182

talking where people are sitting in the car we're talking round tables we're talking you know all of the stuff

183

that counts as hard evidence

184

and i created this absolute beauty of a spreadsheet i know

185

you sent me the csv file yeah you logged everything yeah oh it's

186

so good and then i crunched all the numbers to see

187

if there were you know if any of it would be any use to see

188

if there were statistically significant patterns in the way that traitors behaved

189

and it turns out there was some absolute gems in there um don't like to brag you guys

190

but my burn vote theory was uh came out pretty strong

191

went through all of the first three round tables for every single one of the uk series

192

that had already existed in total those first three three round tables there were 205 votes

193

that were made 34 of from traitors 171 of them from faithful

194

and 33 of them were burn votes eight of those came from traitors

195

and 25 of them came from faithful

196

which kind of shows you the problem right the traitors are doing something

197

that is deeply suspicious but the faithful are such an absolute mess

198

that you can't spot that signal very clearly however because there are

199

so many more faithful than there are traitors as a percentage this was eight out of 34 votes

200

which is 23 .5 percent which means

201

that a quarter of traitors are making a burn vote in

202

the first three round tables right that's that's a massive signal

203

there the faithful on the other hand this 25 was out of 171

204

which is 14 .6 so this is like this is the extent of the signals

205

that you're getting one in four traitors are going to make a burn vote

206

and only one in seven of the faithful are going to do it and in In fact, on these two, if you do the ratio of those two,

207

23 .5 over 14 .6, give you 1 .6.

208

That number essentially tells you how strong this is as a signal.

209

If it's really, really close to 1, then it's as likely to have come from a traitor as it's from a faithful.

210

It doesn't give you any information.

211

If this number starts getting really small, like 0 .0 or whatever, then it's much more likely to have come from a faithful than a traitor

212

and it's a really good indication that you've got somebody in front of you that you can trust.

213

But the bigger this number gets, the more likely it is that you are standing in front of a traitor.

214

So here is the idea to find these numbers.

215

You go back through the data of everything that has happened in previous series.

216

You think of everything that might be even remotely traitorous and you calculate this number for that action.

217

Some of my ideas for things that might be signals, unfortunately, turned out to be worthless because, and I'll say this again, the faithful are a complete mess.

218

Maybe the traitors are voting in the majority, no more than the faithful, that one doesn't work.

219

Maybe they're changing their vote between rounds more often, no more than the faithful, because faithful are all over the place.

220

Maybe they're voting in pairs with another player, no more than the faithful, the faithful are like sheep themselves.

221

But I did find a few that looked genuinely useful.

222

And then I hit another problem.

223

Now, keeping track of it I mean yeah it's gonna be

224

let's talk logistics yes are you allowed to walk around with

225

a notebook like what's your here's the problem I can't bring

226

in a phone laptop even a calculator to the castle

227

so how on earth am I gonna manage to calculate these probabilities as I go I can't do

228

that mass in my head you guys mathematicians not good at mental arithmetic

229

so i'm going to need a shortcut

230

but luckily i can steal from turing again

231

because he had the same problem at bletchley um turing didn't have modern computers

232

because well he hadn't invented them yet um and

233

so he came up with this way to turn all of

234

these really complicated probabilities into basically a point scoring system uh it's very mathematically rigorous

235

and it's just um it requires no more than just adding things up as you go

236

so obviously i stole the same trick you're not just picking

237

these points from random you are calculating them precisely based on

238

the historical data thank you very much alan turing um

239

so here are the points as i work them out from the data

240

uh right one if somebody makes a burn vote that earns them plus 0 .47 points

241

doesn't sound that much but you're looking at tiny little signals here but one of the strongest moments, one of the moments where you get the most information is when the faithfuls manage to catch a traitor,

242

because that reveals not just a really strong signal about the person who's got a traitor, but it also reveals stuff about everybody else who's sitting around the table.

243

Shall I show you what I'm about?

244

Do the traitors who are sitting around the table, who are safe that night, do they pile on?

245

And the answer is absolutely yes they do.

246

If somebody votes for a traitor on the night

247

that a traitor goes down that gets them plus 0 .33 points

248

and the even stronger signal is do they pile on despite having never voted for

249

that traitor before great absolutely they get plus 0 .4 points

250

now here's the thing right your traitors are going to be hiding in

251

that group but the problem is that there's loads of faithful hiding in

252

that group as well which makes it really hard to spot them okay

253

so these are all the really strong traitor signals

254

but it turns out

255

when you look at the data the strongest signals of all are in the opposite direction towards people being faithful

256

so you can account for that by taking points away

257

when a traitor goes down yep

258

if you look at the people who did not pile on they are faithful

259

if the traitor is definitely going down everyone's voting for them

260

only the faithful are stupid enough to stick their head above the parapet

261

and say that they're still innocent so you get minus 1 .28 points

262

if you do not vote for a traitor as they are banished next up

263

if a traitor is banished and you look back at who they voted for historically

264

if somebody's name appears on that list that is a really good signal that they are faithful.

265

So you get minus 1 .69.

266

What a whopper.

267

The biggest one of all is actually one that the internet sort of found.

268

In the early versions of the show people noticed that the last group in for breakfast were always faithful.

269

So production started chucking in the occasional traitor in the last

270

group for breakfast however they only ever chuck in one so

271

if you were in the last group for breakfast with somebody who is then revealed to be a traitor

272

that is the strongest possible signal that you are a faithful look it gets convoluted i'm not not lying

273

that is minus 1 .73

274

so it's not much to go on but it's something so here is my idea going in i have everybody's names

275

hidden away somewhere as the game goes on

276

and i keep this running tally of how many points everybody has

277

and the people who have a bigger number are more likely to be a traitor

278

and those have got a smaller number are more likely to be a faithful

279

and then all you've got to do is you've got to combine those numbers with the potential universes

280

that you've been going through and eliminating

281

and hopefully the traitor identities are going to reveal themselves um that's at least in theory,

282

although you've got to add to that fact

283

that I've got to sort of convince people

284

that I've worked all of this out without looking so suspicious that they either banished me or murdered me.

285

This is going to be quite a difficult ask.

286

This could be the most embarrassing, humbling experience.

287

I could be murdered within four minutes.

288

Do you know how many clothes I've packed?

289

Oh my god.

290

So here is hoping that all of this work wasn't in vain

291

and you know stuff works out for me as well as

292

it did for uh you know the enigma machine with turing um

293

although i admit this one does have a slightly smaller global impact all right

294

if that's the plan i also want to crunch these numbers

295

as we go through the show i'm going to be doing it live in the castle i need someone

296

who can run them alongside it.

297

You called?

298

I'm so excited.

299

So once it's all finished, you can go to Matt's channel where he's going to run the numbers and see whether it actually worked or not.

300

And you can also watch the series itself where you can see me trying to do this in real time.

301

I can't wait.

302

Yeah.

303

And I will report back with how well it went.

304

I believe in you.

305

Okay.

306

I'm down.

307

I'll be honest.

308

Thank you.

📺 Zelfde Kanaal

✨ Aanbevolen Video

53 zinnen

The epic opening scene of Lie To Me

Business EnglishDaily ConversationUniversity Students

260 zinnen

AI Just Crossed the Terrifying Line - Now What?

Business EnglishDaily ConversationUniversity Students

17 zinnen

🔥They Studied Your Weaknesses

Business EnglishUniversity StudentsWorking Adults

629 zinnen

Last Cheater Standing Wins $10,000!

Business EnglishUniversity StudentsWorking Adults

118 zinnen

The Anatomy of Fraud

Business EnglishUniversity StudentsWorking Adults

79 zinnen

3 game theory tactics, explained

Business EnglishUniversity StudentsWorking Adults

41 zinnen

RE2 Unit 11B

IELTS PrepTOEFL PrepBusiness English

182 zinnen

We Let $200 GPT-Pro Think for 7 Days on Unsolved Math

Business EnglishUniversity StudentsWorking Adults

138 zinnen

Statistical Thinking in Science: Crash Course Scientific Thinking #2

Business EnglishUniversity StudentsWorking Adults

190 zinnen

Infiltrating BANK SCAMMERS

Business EnglishUniversity StudentsWorking Adults

Woordenschat en spreektips bij deze les

Deze video bevat 308 zinnen en 3669 woorden om na te spreken. Het gesproken deel duurt 21:48. De spreker praat in een natuurlijk tempo van ongeveer 168 woorden per minuut, dicht bij een alledaags gesprek. 83% van de woorden hoort bij de 3.000 meest gebruikte Engelse woorden; de rest kun je beter vooraf bekijken.

Belangrijke woorden in deze video

15 woorden uit de video die het leren waard zijn, met uitspraak en betekenis:

Woord	Uitspraak	Betekenis
traitor zelfstandig naamwoord	/ˈtɹeɪtɚ/	verrader
faithful bijvoeglijk naamwoord	/ˈfeɪθ.fl̩/	getrouw
castle zelfstandig naamwoord	/ˈkɑːsəl/	kasteel, fort
eliminate werkwoord	/ɪˈlɪm.ɪ.neɪt/	elimineren
reveal werkwoord	/ɹɪˈviːl/	onthullen, zich ontpoppen
essentially bijwoord	/ɪˈsɛnʃəli/	essentieel
steal werkwoord	/stiːl/	stelen, ontvreemden
delete werkwoord	/dɪˈliːt/	verwijderen
pile zelfstandig naamwoord	/paɪl/	stapel
suspicious bijvoeglijk naamwoord	/səˈspɪʃ.əs/	verdacht
predict werkwoord	/pɹɪˈdɪkt/	voorspellen
probability zelfstandig naamwoord	/ˌpɹɑ.bəˈbɪl.ə.ti/	waarschijnlijkheid
minus	/ˈmaɪ.nəs/	min
calculate werkwoord	/ˈkælkjʊleɪt/	berekenen, uitwerken
statistically bijwoord	/stəˈtɪs.tɪ.kə.li/	statistisch gezien

Phrasal verbs die je zult horen

Woord	Uitspraak	Betekenis
work out werkwoord	/ˌwɝk ˈaʊt/	uitrekenen
go down werkwoord		afdalen
come up with werkwoord		bedenken, verzinnen
go back werkwoord		teruggaan
turn out werkwoord		uitvallen, resulteren
come out werkwoord	/ˌkʌm ˈaʊt/	uitkomen
look up werkwoord	/ˌlʊk ˈʌp/	opkijken

Grammatica in deze video

De structuren die de spreker het meest gebruikt, met de exacte woorden uit de video:

Structuur	In de video
Present perfect continuous have/has been + -ing — iets dat eerder begon en nog bezig is	i've been doing · you've been going
Present perfect have/has + voltooid deelwoord — iets uit het verleden dat nu nog telt	they've ditched · have played · i've done
Lijdende vorm be + voltooid deelwoord — het gaat om wat er gebeurt, niet om wie het doet	might be tempted · are dilated · be picked
Betrekkelijke bijzinnen who / which + zin — extra informatie over een persoon of ding	people who run · percent which means · well which makes

Uitspraak om op te letten

De spreker gebruikt 105 samentrekkingen en verkorte vormen, zoals don't, I'm, I've. Spreek ze kort uit, zoals je ze hoort.

De “th”-klanken: faithful /ˈfeɪθ.fl̩/, bother /ˈbɑðɚ/, leather /ˈlɛðɚ/, mathematics /mæθ(.ə)ˈmæt.ɪks/, mathematical /ˌmæθ(.ə)ˈmæt.ɪ.kəl/
De klanken “sh” en “zh”: essentially /ɪˈsɛnʃəli/, suspicious /səˈspɪʃ.əs/, indication /ɪndɪˈkeɪʃən/, combinations /kɑmbɪˈneɪʃənz/, temptation /ˌtɛmpˈteɪʃən/
Lange woorden — let op de klemtoon: eliminate /ɪˈlɪm.ɪ.neɪt/, essentially /ɪˈsɛnʃəli/, probability /ˌpɹɑ.bəˈbɪl.ə.ti/, statistically /stəˈtɪs.tɪ.kə.li/, psychology /saɪˈkɑ.lə.d͡ʒi/

Zo oefen je met deze video

Luister de hele video één keer zonder te spreken en noteer de woorden die je niet kent.
Begin op 0,75× snelheid, spreek zin voor zin na en ga terug naar normale snelheid zodra het makkelijk gaat.
Neem jezelf op en vergelijk met het origineel; let daarbij op woorden als traitor, faithful, castle.

Wat is de Shadowing-techniek?

Shadowing is een wetenschappelijk onderbouwde taalleermethode die oorspronkelijk is ontwikkeld voor professionele tolkentraining en gepopulariseerd door polyglot Dr. Alexander Arguelles. De methode is eenvoudig maar krachtig: je luistert naar native Engelse audio en herhaalt het onmiddellijk hardop — als een schaduw die de spreker volgt met slechts 1–2 seconden vertraging. In tegenstelling tot passief luisteren of grammaticadrills, dwingt shadowing je hersenen en mondspieren om echte spraakpatronen tegelijkertijd te verwerken en te reproduceren. Onderzoek toont aan dat het de uitspraaknauwkeurigheid, intonatie, ritme, verbonden spraak, luisterbegrip en spreekvaardigheid aanzienlijk verbetert — waardoor het een van de meest effectieve methoden is voor IELTS Speaking-voorbereiding en echte Engelse communicatie.

Shadowing-techniek: lees de volledige stap-voor-stap-gids →

Shadowing Practice: How would Alan Turing have played the Traitors? - Learn English Speaking with Video

I'm ever so pleased to make your acquaintance

An unfortunate naming accident

Why birds don't crash (and what humans can learn from them)

Keep Calm and Carrot On - The lie about carrots and night vision

The epic opening scene of Lie To Me

AI Just Crossed the Terrifying Line - Now What?

🔥They Studied Your Weaknesses

Last Cheater Standing Wins $10,000!

The Anatomy of Fraud

3 game theory tactics, explained

RE2 Unit 11B

We Let $200 GPT-Pro Think for 7 Days on Unsolved Math

Statistical Thinking in Science: Crash Course Scientific Thinking #2

Infiltrating BANK SCAMMERS

Woordenschat en spreektips bij deze les

Belangrijke woorden in deze video

Phrasal verbs die je zult horen

Grammatica in deze video

Uitspraak om op te letten

Zo oefen je met deze video

Wat is de Shadowing-techniek?