From my perspective - and I know it is a particular one - there has not been much to be proud about over the last decade.
We can and should have been better than we were. It does the City no favours to give it a free pass from the requirement to apply some judgment and ask itself not just whether it can do something but whether it should.
No-one is suggesting that the City should be given a free pass; humans are fallible, and laziness, complacency, greed, and occasionally outright dishonesty are inevitably a part of that. That is true of the City, but it's also true of the car industry, estate agents, construction, and, yes, local councils, the NHS and educational establishments as well. Fiddling the waiting-list figures to meet NHS targets is morally no different to fiddling the LIBOR rate, is it? We just have to accept that these things may happen in any human activity, and set up our regulatory framework, and our training and recruitment and management processes, to minimise them. We don't have to throw out the baby with the bathwater.
Mr. Jessop, but there'd be no Prost-Senna, Schumacher-Hill, Hamilton-Rosberg rivalries. That's a bit part of the draw to F1. People like the needle between drivers.
@PolhomeEditor: BREAKING: Jamie Reed has tendered his resignation as Copeland MP, a couple of weeks earlier than planned. Tristram Hunt quits today too.
@PolhomeEditor: BREAKING: Jamie Reed has tendered his resignation as Copeland MP, a couple of weeks earlier than planned. Tristram Hunt quits today too.
Great stuff, that time value of Labour value can work itself through in Copeland and Stoke now.
Mr. Pulpstar, there was chatter a little time ago about a formula with self-driven cars. I think it was Heidfeld who said it'd be tedious because there'd be no element of human drama, or danger.
I have a very nice car and I enjoy driving. I am a hopeless passenger and would be bored witless just sitting in a car driven by someone else. Also, practically the last time I was a passenger I was thrown at the windshield and ended up in a Nairobi hospital being operated on. It has rather ruined the passenger experience for me. I'm not really willing to trust other drivers, whether human or robot.
Also I quite like shouting at red lights.
Unlike in a train you can't read in a car. So what are you supposed to do?
I prefer a car and chauffeur. All stress free of riding the train first class without having to do deal with rowdy peasants who sneak into first class
I was once late for a meeting with a friend of mine because there were problems on the Tube.
He sympathised with me: "Public transport is horrible, isn't it. I sacked my pilot yesterday for shouting at air traffic control - and it had to be when the chopper was in for a service. Would you believe that I had to fly BA?"
Don't passive funds outperform active ones in the long run?
The average actively managed fund will lose you about 2% a year compared to a passive one, yes. The fund manager will lose you 1% a year through their stock picks, and you'll pay an additional 1% in fees.
So, you're basically paying up to get performance that's worse than a chimpanzee.
A week today I am attending 4 presentations by fund managers who want to actively manage a pension fund I am a trustee for. Are you saying I am wasting my time?
You may want to match the liabilities with pension maturity to lock in the fund at a known cost. Some pension fund managers may be able to do this more cheaply than others. Academic studies have shown the best lon term fund managers are those with the lowest cost (or those with inside information who use it illegally).
Banks who want to sell wholesale products from London to EU countries can continue to do so outside the EU as New York, Zurich and Singapore currently do.
Unless remaining EU countries ban their companies from using London banks (illegal under WTO?), then EU compnaies will continue to acquire finance from London banks because it is the biggest and best market.
UK Banks who want to sell retail products to personal customers in EU countries will not be able to do so from branches set up in EU countries post Brexit without either equivalence, passporting or using an EU subsidiary. However, to have any significant presence in an EU country, UK banks have found they need to acquire a local bank with a local brand name - as HSBC did when they acquired Midland Bank in the UK.
The only significant impact on banking from moving out of the EU is if EU clearing regulators act politically and make it a reglautory requirement about where transaction clearing is done. Even so it may be done in 'the cloud' and hard to establish which geographical country the cloud resides at any moment in time. especially if blockchain technology takes off.
Clearing (and more importantly netting) of trades is a bit more complex than that, because clearing houses need to be implicitly or explicitly backed by central banks. Given potentially unlimited exposure, they are - in effect - guaranteed by their banks.
Would the UK government step up to bail out Euroclear if it were to go bust leaving those with Sterling contracts out of pocket? Likewise, would the ECB/EU step in if it was Euro denominated deals at the LCH?
Clearing (and more importantly netting) of trades is a bit more complex than that, because clearing houses need to be implicitly or explicitly backed by central banks. Given potentially unlimited exposure, they are - in effect - guaranteed by their banks.
Would the UK government step up to bail out Euroclear if it were to go bust leaving those with Sterling contracts out of pocket? Likewise, would the ECB/EU step in if it was Euro denominated deals at the LCH?
You are being exceedingly polite about a very ignorant post!
It'd be interesting to see what a proper AI program could do round Monza compared to say Lewis Hamilton.
Probably traffic dependent. I suspect a machine could record a faster timed lap, but might not win a race (yet)
On an empty track a driverless car could be programmed (not trained) to beat the best drivers using current tech. They may be able to make them do so whilst other drivers are on the track (though I doubt they'd try that at the moment for safety reasons).
If it ever got to the state where they could, it would be interesting to see how quickly the drivers would pick up on the flaws in the programming and learn how to beat the automation. And how quickly the programmers would be able to reprogram the machines to cover those flaws.
Until, that is, true machine learning comes in.
True machine learning is already here though. I know it might seem like just a board game, but "Go" to a decent level isn't solvable simply through brute force tree analysis in the way chess is. Combine that with the progress various Cali-tech firms are making on self driving cars in real traffic which fundamentally is a harder problem to 'solve' than any F1 track and I think a fully automated F1 race winning car is eminently possible.
"True machine learning is already here though."
It's not though, or at least in depth. Go, like Chess, has some fairly simple rules that are well defined. It's incredibly shallow machine learning. Try defining similar rules to act as bounds for (say) driving: it's massively more complex problem than just inputting the highway code.
The vast majority of AI mentioned in the press is just smoke and mirrors. People are (justifiably) sceptical about media reports on the economy or politics, yet swallow any old guff from a journalist when it comes to technology.
As a related aside, I wonder how the Go versus human games work on different size boards? I wonder if a larger or smaller board makes the games easier or harder for the machine to win.
Out of curiosity, how much did the trade unions contribute to the government in tax revenues and how does it compare to the banking and financial services industry in 2017?
I'm going to go out on a limb and say it wasn't the trade unions that contributed the most.
The large banks are able to borrow almost interest free at the moment aren't they ?
Very easy to make money with that model.
It is not the level of interest rates which matter but
a) the interest margin between borrowing and lending rates and
b) how much of the orginal loan gets paid back over the loan period.
It'd be interesting to see what a proper AI program could do round Monza compared to say Lewis Hamilton.
Probably traffic dependent. I suspect a machine could record a faster timed lap, but might not win a race (yet)
On an empty track a driverless car could be programmed (not trained) to beat the best drivers using current tech. They may be able to make them do so whilst other drivers are on the track (though I doubt they'd try that at the moment for safety reasons).
If it ever got to the state where they could, it would be interesting to see how quickly the drivers would pick up on the flaws in the programming and learn how to beat the automation. And how quickly the programmers would be able to reprogram the machines to cover those flaws.
Until, that is, true machine learning comes in.
I would expect AI to be programmed to be somewhat defensive and thus far more bullyable than the average F1 driver.
Mr. Jessop, but there'd be no Prost-Senna, Schumacher-Hill, Hamilton-Rosberg rivalries. That's a bit part of the draw to F1. People like the needle between drivers.
Well, indeed. Automated Grand Prix comes, I think, under the heading of: Things We Could Do But Shouldn't Do.
P.S. I accidentally clicked (and then unclicked) the "Off Topic" instead of the "Quote" button. I hope the system doesn't hold it against you. What purpose does the "Off Topic" button serve here anyway?
It'd be interesting to see what a proper AI program could do round Monza compared to say Lewis Hamilton.
Probably traffic dependent. I suspect a machine could record a faster timed lap, but might not win a race (yet)
On an empty track a driverless car could be programmed (not trained) to beat the best drivers using current tech. They may be able to make them do so whilst other drivers are on the track (though I doubt they'd try that at the moment for safety reasons).
If it ever got to the state where they could, it would be interesting to see how quickly the drivers would pick up on the flaws in the programming and learn how to beat the automation. And how quickly the programmers would be able to reprogram the machines to cover those flaws.
Until, that is, true machine learning comes in.
True machine learning is already here though. I know it might seem like just a board game, but "Go" to a decent level isn't solvable simply through brute force tree analysis in the way chess is. Combine that with the progress various Cali-tech firms are making on self driving cars in real traffic which fundamentally is a harder problem to 'solve' than any F1 track and I think a fully automated F1 race winning car is eminently possible.
"True machine learning is already here though."
It's not though, or at least in depth. Go, like Chess, has some fairly simple rules that are well defined. It's incredibly shallow machine learning. Try defining similar rules to act as bounds for (say) driving: it's massively more complex problem than just inputting the highway code.
The vast majority of AI mentioned in the press is just smoke and mirrors. People are (justifiably) sceptical about media reports on the economy or politics, yet swallow any old guff from a journalist when it comes to technology.
As a related aside, I wonder how the Go versus human games work on different size boards? I wonder if a larger or smaller board makes the games easier or harder for the machine to win.
What's the view on the stockmarket next week? If, as expected, the SC declares a vote is needed and a bill is passed in HoC, would we expect a 'correction'? Alternatively, it's already priced in
Next week's movements on the stockmarket are not knowable. It's a random walk.
Holding by-eelctions together splits Labour's resources, but prevents any feedback effect. Oddly I think that makes them more likely to hold Stoke and less likely to hold Copeland. Stoke is a lot easier to get to...
Those figures look like a load of hot air anyway. Nobody really knows the limits of automation. All I would say is that the application of robotics, AI, machine learning and the like is going to be wide, and probably wider than most people think.
AI will inevitably lead to a universal basic income in a decade or two, Finland is already experimenting with it, as work becomes more flexible and less full time and permanent. People will also need time to retrain and tax will shift more to capital
Sometimes whilst I'm driving I try to analyse everything I'm doing: not just my outputs to the controls, but also everything I'm absorbing from the environment around me. Even noise can be an important input.
It's very noticeable how much less input I have to process when driving on a quiet dual carriageway compared to going through the centre of Cambridge in dusk at rush hour in the rain.
This is why driving on a dual carriageway can be so less stressful than driving through a city: you are doing much less work. It's also why they've been able to make good driverless cars on freeways / motorways long before they've got towns and cities sorted.
Holding by-eelctions together splits Labour's resources, but prevents any feedback effect. Oddly I think that makes them more likely to hold Stoke and less likely to hold Copeland. Stoke is a lot easier to get to...
My favourite bet is still the £20 I have on the double at 31-10.
I'm glad they've moved the writ quickly at any rate.
Holding by-eelctions together splits Labour's resources, but prevents any feedback effect. Oddly I think that makes them more likely to hold Stoke and less likely to hold Copeland. Stoke is a lot easier to get to...
My favourite bet is still the £20 I have on the double at 31-10.
Isn't Labour in Copeland at 2.5 (40%) and 1.8 (55%) in Stoke at the moment? Perhaps my maths is garbled.
From my perspective - and I know it is a particular one - there has not been much to be proud about over the last decade.
We can and should have been better than we were. It does the City no favours to give it a free pass from the requirement to apply some judgment and ask itself not just whether it can do something but whether it should.
No-one is suggesting that the City should be given a free pass; humans are fallible, and laziness, complacency, greed, and occasionally outright dishonesty are inevitably a part of that. That is true of the City, but it's also true of the car industry, estate agents, construction, and, yes, local councils, the NHS and educational establishments as well. Fiddling the waiting-list figures to meet NHS targets is morally no different to fiddling the LIBOR rate, is it? We just have to accept that these things may happen in any human activity, and set up our regulatory framework, and our training and recruitment and management processes, to minimise them. We don't have to throw out the baby with the bathwater.
We need to reintroduce hazard to banking. Their staff bet big. Win and the the heavens rain money. Lose and they move to the next bank along. The risk / reward profile almost enforces a 'for myself' approach, not a service approach. Would separating the utility bit from the casino bit deliver? Maybe. Should we be more upfront with customers about the fact their money is not 'in the bank' and their life savings are one bank run away from not being there at all? For sure. Are corporate governance bodies failing? Absolutely. Greater personal implications for bosses of failed spiv banks? Faster please. And make sure the regualtory environment manages systemic risk properly. Gordon Brown screwed that and I'm not we've since properly fixed it.
It'd be interesting to see what a proper AI program could do round Monza compared to say Lewis Hamilton.
Probably traffic dependent. I suspect a machine could record a faster timed lap, but might not win a race (yet)
On an empty track a driverless car could be programmed (not trained) to beat the best drivers using current tech. They may be able to make them do so whilst other drivers are on the track (though I doubt they'd try that at the moment for safety reasons).
If it ever got to the state where they could, it would be interesting to see how quickly the drivers would pick up on the flaws in the programming and learn how to beat the automation. And how quickly the programmers would be able to reprogram the machines to cover those flaws.
Until, that is, true machine learning comes in.
True machine learning is already here though. I know it might seem like just a board game, but "Go" to a decent level isn't solvable simply through brute force tree analysis in the way chess is. Combine that with the progress various Cali-tech firms are making on self driving cars in real traffic which fundamentally is a harder problem to 'solve' than any F1 track and I think a fully automated F1 race winning car is eminently possible.
"True machine learning is already here though."
It's not though, or at least in depth. Go, like Chess, has some fairly simple rules that are well defined. It's incredibly shallow machine learning. Try defining similar rules to act as bounds for (say) driving: it's massively more complex problem than just inputting the highway code.
The vast majority of AI mentioned in the press is just smoke and mirrors. People are (justifiably) sceptical about media reports on the economy or politics, yet swallow any old guff from a journalist when it comes to technology.
As a related aside, I wonder how the Go versus human games work on different size boards? I wonder if a larger or smaller board makes the games easier or harder for the machine to win.
The larger the board, the bigger DeepMind's advantage.
Thanks; that's interesting.
I'll be more impressed when a program coded to beat someone at chess can beat someone else at go, with no programming except knowing the rules. Oh, and then writ a novel.
Deepmind are doing some interesting stuff, though I'm wary about them getting access to medical records.
It's interesting because it starts logically enough but the logic leads inexorably to a completely nutty conclusion. He correctly in my view puts his finger on a key point (his para 6), which is that any successful attempt by parliament to put restrictions on the PM's negotiating position would inevitably mean we get a less favourable deal from the EU.
After that, he descends into an amusingly absurd fantasy.
It all serves to show that Labour really are in a bind about this. The LibDems perhaps less so, because they are not pretending to be a party of government.
Another thought on the 'banks will move' debate: It's more likely that we should be having a 'banks will get disintermdiated' debate. They've protected their business model for so long that they seem to have fialed to spot inherent weaknesses. Weaknesses that fintech is rapidly exploiting. If you want t: 1. Borrow 2. Raise equity 3. Exchange forex 4. Invest 5. Blah blah blah there are an ever increasing range of options for customers that don't involve a bank at all.
The City should be MUCH more concerned about its business model than its location. And we should all push to make London the absolute centre of European fintech. Oh...it is. :-)
Holding by-eelctions together splits Labour's resources, but prevents any feedback effect. Oddly I think that makes them more likely to hold Stoke and less likely to hold Copeland. Stoke is a lot easier to get to...
My favourite bet is still the £20 I have on the double at 31-10.
Isn't Labour in Copeland at 2.5 (40%) and 1.8 (55%) in Stoke at the moment? Perhaps my maths is garbled.
Well I've got "proper" books on them both but the correlation surely pays for the 7-2/31-10 'gap' ?
Another thought on the 'banks will move' debate: It's more likely that we should be having a 'banks will get disintermdiated' debate. They've protected their business model for so long that they seem to have fialed to spot inherent weaknesses. Weaknesses that fintech is rapidly exploiting. If you want t: 1. Borrow 2. Raise equity 3. Exchange forex 4. Invest 5. Blah blah blah there are an ever increasing range of options for customers that don't involve a bank at all.
The City should be MUCH more concerned about its business model than its location. And we should all push to make London the absolute centre of European fintech. Oh...it is. :-)
And yet when it comes to digital intermediation - another big macro trend - it'a a corporation in California picking up the lion's share of the new business in something as 'London' as catching a ride.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
Perhaps the UK could also approach Canada for political union - her majesty might have to move to Ottawa but it would save us having to go through the costly and unpredictable process of Brexit as we would be able to join the Canadian trade agreement with the EU.
Perhaps the UK could also approach Canada for political union - her majesty might have to move to Ottawa but it would save us having to go through the costly and unpredictable process of Brexit as we would be able to join the Canadian trade agreement with the EU.
Don't passive funds outperform active ones in the long run?
The average actively managed fund will lose you about 2% a year compared to a passive one, yes. The fund manager will lose you 1% a year through their stock picks, and you'll pay an additional 1% in fees.
So, you're basically paying up to get performance that's worse than a chimpanzee.
A week today I am attending 4 presentations by fund managers who want to actively manage a pension fund I am a trustee for. Are you saying I am wasting my time?
David , hopefully you will be giving us a full report on your presentations. I am currently looking at where to put my pension fund. Need to get a SIPP and where to invest.
Perhaps the UK could also approach Canada for political union - her majesty might have to move to Ottawa but it would save us having to go through the costly and unpredictable process of Brexit as we would be able to join the Canadian trade agreement with the EU.
Brilliant.
The Canzuk lobby want to move the capital to Canada anyway so it fits with the plan.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
No, not at all, because unlike chess or go it is a hidden information game & with no limit there is an infinite number of bet sizes in each betting round. Academics have been at this for longer than chess, and no suggestion they are close on poker games including more than 2 players or more complex variants such as pot limit Omaha.
Don't passive funds outperform active ones in the long run?
The average actively managed fund will lose you about 2% a year compared to a passive one, yes. The fund manager will lose you 1% a year through their stock picks, and you'll pay an additional 1% in fees.
So, you're basically paying up to get performance that's worse than a chimpanzee.
A week today I am attending 4 presentations by fund managers who want to actively manage a pension fund I am a trustee for. Are you saying I am wasting my time?
David , hopefully you will be giving us a full report on your presentations. I am currently looking at where to put my pension fund. Need to get a SIPP and where to invest.
It'd be interesting to see what a proper AI program could do round Monza compared to say Lewis Hamilton.
Probably traffic dependent. I suspect a machine could record a faster timed lap, but might not win a race (yet)
On an empty track a driverless car could be programmed (not trained) to beat the best drivers using current tech. They may be able to make them do so whilst other drivers are on the track (though I doubt they'd try that at the moment for safety reasons).
If it ever got to the state where they could, it would be interesting to see how quickly the drivers would pick up on the flaws in the programming and learn how to beat the automation. And how quickly the programmers would be able to reprogram the machines to cover those flaws.
Until, that is, true machine learning comes in.
True machine learning is already here though. I know it might seem like just a board game, but "Go" to a decent level isn't solvable simply through brute force tree analysis in the way chess is. Combine that with the progress various Cali-tech firms are making on self driving cars in real traffic which fundamentally is a harder problem to 'solve' than any F1 track and I think a fully automated F1 race winning car is eminently possible.
Getting an AI racing car around a track quickly is relatively straightforward, has already been demonstrated and there are plans for a race series of them to help develop the competitive instinct, that last few percent where races are won and lost.
An autonomous F1 car would be very uncompaetitive against real racing drivers, the AI would be way too conservative with things like yellow or blue flags, and would struggle to overtake in situations where the other driver has to be co-operative. In F1 specifically it would also struggle with race strategy and tyre management without a team of humans advising it as happens with a real driver.
A developed-from-scratch autonomous racing car will be very quick on the brakes and around the corners, not having to worry about getting a stiff neck after a couple of hours' racing.
But let's face it, a huge draw of the F1 race is the chance of a big accident, and seeing someone either walk away or end up in the hospital. It's one of the few things left which is popular because it's bloody dangerous - even if the safety record is massively improved in recent times.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
(emphasis added)
I hope you don't risk real money at poker thinking you can beat what you know to be a perfectly-playing opponent.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
If a human bets big its normally best to fold too.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
This is completely wrong. There is a game theory optimal solution ie a Nash equilibrium to HU poker. All solutions look to find this solution or as close as possible to.
Such a solution does not need to consider the opponent's strategy (or adjust) & we know that the GTO solition is a mixed strategy ie you don't just make a certain move in a particular scenario , rather it is about the proportions you take each possible action ie raise, call, fold when facing a bet.
2-3 years ago a basically near perfect GTO solition for limit HU poker was found. NL is still a long way off, but it seems to approximated GTO solution is looking better than some of the best humans despite being miles away from GTO ie humans were no were near as good as they thought.
From my perspective - and I know it is a particular one - there has not been much to be proud about over the last decade.
We can and should have been better than we were. It does the City no favours to give it a free pass from the requirement to apply some judgment and ask itself not just whether it can do something but whether it should.
No-one is suggesting that the City should be given a free pass; humans are fallible, and laziness, complacency, greed, and occasionally outright dishonesty are inevitably a part of that. That is true of the City, but it's also true of the car industry, estate agents, construction, and, yes, local councils, the NHS and educational establishments as well. Fiddling the waiting-list figures to meet NHS targets is morally no different to fiddling the LIBOR rate, is it? We just have to accept that these things may happen in any human activity, and set up our regulatory framework, and our training and recruitment and management processes, to minimise them. We don't have to throw out the baby with the bathwater.
"Fiddling the waiting-list figures to meet NHS targets is morally no different to fiddling the LIBOR rate, is it?"
One was criminal. The other isn't.
Oh dear! A lot of whataboutery from you. I'm a bit surprised.
Of course bad behavior exists in other organisations: look at MPs for instance and the police. But I was writing about the City and pointing out the gap between its high opinion of itself, the reality and what that meant for public policy at a time when the City is under pressure and could do with some friends.
Its past behavior and willingness to excuse itself on the grounds of its revenues is not a sufficient answer anymore. Not that it should ever have been, IMO. The City is learning the hard way now the opposite of the Boy Crying Wolf lesson i.e. if you spend all your time saying how wonderful you are and then it turns out that you're not wonderful, don't be surprised when people don't listen to you when you say how important you are to them.
For an industry which has a lot of clever people working in it, it is remarkable how few intelligent - and emotionally intelligent - spokespeople it has. I cringe when I hear someone speaking for the City. They invariably get the tone (as well as the content) wrong. May is not listening to the City because she is listening to others. But she is also not listening to the City because the City's case is all too often presented with a mix of bluster and arrogance and "don't you know who I am" which grates on those who do not see it the way the City would like to be seen.
Christ, if it grates on me - and I've more experience than most of it - imagine how it grates on people in Barrow earning a pittance by comparison.
Of course CA has always had a soft spot for shysters and mountebanks. Perhaps Nige can found his own religion.
Nige as the new Jim Jones?
Farage would draw the line at buggering his male followers. Probably.
Given the level of his access to the man who will become as close as it gets to being King of the World at 5 p.m. our time, I don't think saying "LOL what a bellend" (which he is, of course) really does justice to his political importance. He gets in to the history books anyway, because if you look at the utterly unbelievable uselessness of everyone in UKIP, except possibly Carswell who was too late to the party to count, Cameron might never have been pressurised into a refendum.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
(emphasis added)
I hope you don't risk real money at poker thinking you can beat what you know to be a perfectly-playing opponent.
Only ever play poker with friends for beer money, as an alternative to an expensive night in a bar.
It's one of those games that's simple enough on the face of it, but those who do it professionally are *really* good at it.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
If a human bets big its normally best to fold too.
LOL true, but we all enjoy that time when we call someone's bluff and he's got nothing!
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
(emphasis added)
I hope you don't risk real money at poker thinking you can beat what you know to be a perfectly-playing opponent.
Only ever play poker with friends for beer money, as an alternative to an expensive night in a bar.
It's one of those games that's simple enough on the face of it, but those who do it professionally are *really* good at it.
Interestingly the current test going on is showing the best human players aren't anywhere near as good as thought.
This is very similar to chess. After deep blue, chess players realises they werent anywhere near "perfect". Today's top chess players are miles better because they can see the computer learned stragies and build on them. Apparently computer can beat the beat human at chess, but human with computer assistance beats the computer on its own.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
If a human bets big its normally best to fold too.
LOL true, but we all enjoy that time when we call someone's bluff and he's got nothing!
@ Cyclefree "Maybe – though Brexit is an odd and potentially harsh way to achieve such a rebalancing." In the world of HROs (high reliability organisations) this failure that brings about organisational change is called the 'brutal audit'. Seems appropriate here, too.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
(emphasis added)
I hope you don't risk real money at poker thinking you can beat what you know to be a perfectly-playing opponent.
Only ever play poker with friends for beer money, as an alternative to an expensive night in a bar.
It's one of those games that's simple enough on the face of it, but those who do it professionally are *really* good at it.
It's not really a card game, though, is it? Sure, calculating the probabilities is what all the TV shows do for you. But ultimately it is psychology and micro-expressions.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
(emphasis added)
I hope you don't risk real money at poker thinking you can beat what you know to be a perfectly-playing opponent.
Only ever play poker with friends for beer money, as an alternative to an expensive night in a bar.
It's one of those games that's simple enough on the face of it, but those who do it professionally are *really* good at it.
It's not really a card game, though, is it? Sure, calculating the probabilities is what all the TV shows do for you. But ultimately it is psychology and micro-expressions.
Correct. Which is why the computers can't beat the humans at it - yet!
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
(emphasis added)
I hope you don't risk real money at poker thinking you can beat what you know to be a perfectly-playing opponent.
Only ever play poker with friends for beer money, as an alternative to an expensive night in a bar.
It's one of those games that's simple enough on the face of it, but those who do it professionally are *really* good at it.
It's not really a card game, though, is it? Sure, calculating the probabilities is what all the TV shows do for you. But ultimately it is psychology and micro-expressions.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
(emphasis added)
I hope you don't risk real money at poker thinking you can beat what you know to be a perfectly-playing opponent.
Only ever play poker with friends for beer money, as an alternative to an expensive night in a bar.
It's one of those games that's simple enough on the face of it, but those who do it professionally are *really* good at it.
It's not really a card game, though, is it? Sure, calculating the probabilities is what all the TV shows do for you. But ultimately it is psychology and micro-expressions.
Correct. Which is why the computers can't beat the humans at it - yet!
"Fiddling the waiting-list figures to meet NHS targets is morally no different to fiddling the LIBOR rate, is it?"
One was criminal. The other isn't.
Oh dear! A lot of whataboutery from you. I'm a bit surprised. ...
It's not 'whataboutery' to point out that there is nothing specific to the City about bad behaviour. So why do people pick on the City specifically? The answer is presumably the obvious and ugly one: envy - never a healthy emotion.
Perhaps the UK could also approach Canada for political union - her majesty might have to move to Ottawa but it would save us having to go through the costly and unpredictable process of Brexit as we would be able to join the Canadian trade agreement with the EU.
Brilliant.
There's been a campaign for years for WA, OR and BC to merge to form Cascadia. Which has to be the worst proposed name for a country that I've ever heard.
Perhaps the UK could also approach Canada for political union - her majesty might have to move to Ottawa but it would save us having to go through the costly and unpredictable process of Brexit as we would be able to join the Canadian trade agreement with the EU.
Brilliant.
There's been a campaign for years for WA, OR and BC to merge to form Cascadia. Which has to be the worst proposed name for a country that I've ever heard.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
(emphasis added)
I hope you don't risk real money at poker thinking you can beat what you know to be a perfectly-playing opponent.
Only ever play poker with friends for beer money, as an alternative to an expensive night in a bar.
It's one of those games that's simple enough on the face of it, but those who do it professionally are *really* good at it.
Like backgammon. It is easy to think that it's 66% the fall of the dice and 33% common sense, and if you think that a pro will fleece you.
No-one is suggesting that the City should be given a free pass; humans are fallible, and laziness, complacency, greed, and occasionally outright dishonesty are inevitably a part of that. ... Fiddling the waiting-list figures to meet NHS targets is morally no different to fiddling the LIBOR rate, is it? We just have to accept that these things may happen in any human activity, and set up our regulatory framework, and our training and recruitment and management processes, to minimise them. We don't have to throw out the baby with the bathwater.
"Fiddling the waiting-list figures to meet NHS targets is morally no different to fiddling the LIBOR rate, is it?"
One was criminal. The other isn't.
A lot of whataboutery from you. I'm a bit surprised.
Of course bad behavior exists in other organisations: look at MPs for instance and the police.
Surprisingly, I find myself more on Richard's side on this than yours, Ms Cyclefree. Human behaviour is the result of antecedents and consequences.
Translating what Richard says into what I do, he is essentially saying that the way organizations have set themselves up, they way they measure themselves (e.g. waiting lists for NHS) and the way that they reward people drives most of the behaviour. That is the essence of systems approaches to safety (and excellence).
Whether the something that is wrong (because it leads to suboptimal outcomes) is criminal or not really is arbitrary based on what we choose to legislate. Even criminal behaviour happens in large part because that is what the system sets up some fairly predictable number of people to do.
A statistic I find fascinating from accident causation research is that, where accidents do happen due to 'human error' (which should generally be treated as a symptom of the system, not a root cause), 85% of people put in that situation would have reacted the same way and caused the same accident. I do not know what the statistics are for criminal behaviour, but it would surprise me if that general principle did not translate.
The other aspect of Richard's response I wholeheartedly agree with is that success breeds complacency, arrogance and laziness, and reduces our willingness to listen and learn, or even to notice that things are beginning to go off the tracks unless we consciously set things up to question ourselves despite our success.
Human error is rarely what the name says it is. It is sometimes human performance variance, but even more usually environmental variance. I would expand upon Richard's lists of measures by more explicitly breaking down management processes - organisational structure and culture, performance management systems (in particular) and alignment of goals with ethics and safety values are absolutely key to getting behaviours you want and minimizing behaviours you don't.
Surprisingly, I find myself more on Richard's side on this than yours, Ms Cyclefree. Human behaviour is the result of antecedents and consequences.
Translating what Richard says into what I do, he is essentially saying that the way organizations have set themselves up, they way they measure themselves (e.g. waiting lists for NHS) and the way that they reward people drives most of the behaviour. That is the essence of systems approaches to safety (and excellence).
Whether the something that is wrong (because it leads to suboptimal outcomes) is criminal or not really is arbitrary based on what we choose to legislate. Even criminal behaviour happens in large part because that is what the system sets up some fairly predictable number of people to do.
A statistic I find fascinating from accident causation research is that, where accidents do happen due to 'human error' (which should generally be treated as a symptom of the system, not a root cause), 85% of people put in that situation would have reacted the same way and caused the same accident. I do not know what the statistics are for criminal behaviour, but it would surprise me if that general principle did not translate.
The other aspect of Richard's response I wholeheartedly agree with is that success breeds complacency, arrogance and laziness, and reduces our willingness to listen and learn, or even to notice that things are beginning to go off the tracks unless we consciously set things up to question ourselves despite our success.
Human error is rarely what the name says it is. It is sometimes human performance variance, but even more usually environmental variance. I would expand upon Richard's lists of measures by more explicitly breaking down management processes - organisational structure and culture, performance management systems (in particular) and alignment of goals with ethics and safety values are absolutely key to getting behaviours you want and minimizing behaviours you don't.
"Fiddling the waiting-list figures to meet NHS targets is morally no different to fiddling the LIBOR rate, is it?"
One was criminal. The other isn't.
Oh dear! A lot of whataboutery from you. I'm a bit surprised. ...
It's not 'whataboutery' to point out that there is nothing specific to the City about bad behaviour. So why do people pick on the City specifically? The answer is presumably the obvious and ugly one: envy - never a healthy emotion.
I'll give you two reasons:-
1. The City has cost us a lot of money as well as making us a lot in the past. Money incites emotion and it is not just about envy. Look at peoples' reaction to MPs fiddling their expenses. They didn't get so het up about police misfeasance.
2. Banking is about trust. It's the only thing which matters in banking. That trust has been broken or very severely damaged. And once you lose trust it takes an age to rebuild it. And those who have had their trust abused can feel vitriolic about it. Understandably.
It is far too simplistic (and comforting) to say that it is just envy. If you do you risk IMO missing the importance of the very fundamental failings which recent scandals have shown us and, more alarmingly, you are more likely not to take a good hard look at yourself and really take the steps needed to improve matters. This is hard. And people tend to shy away from hard steps like this. Especially if there are people around to say that the criticisms are all down to envy, blah, blah....
Re: Cyclefree article Are the changes that fin tech is already bringing to commercial SME lending such as P2PFA, going to cause radical changes throughout the sector? With 50% growth pa in P2PFA, the oligopolies of main street banks may look time limited. Note that peer to peer now starting into mortgages albeit at very very small levels. Here is one recent article. http://www.businessinsider.com/p2pfa-q4-2016-stats-strong-end-to-year-p2p-funding-circle-2017-1?r=UK&IR=T
Donald Trump is being inaugurated as President today. I actually predicted he would win the election and yet it still seems somehow surreal that he's actually going to be President. 2016 was just the beginning, I think we will be in for a very interesting next couple of years.
Re: Cyclefree article Are the changes that fin tech is already bringing to commercial SME lending such as P2PFA, going to cause radical changes throughout the sector? With 50% growth pa in P2PFA, the oligopolies of main street banks may look time limited. Note that peer to peer now starting into mortgages albeit at very very small levels. Here is one recent article. http://www.businessinsider.com/p2pfa-q4-2016-stats-strong-end-to-year-p2p-funding-circle-2017-1?r=UK&IR=T
Both myself and Rod formerly of this parish have invested in this one. Less defaults, better security and higher rates than Funding Circle.
The loanbook isn't as mature, availability is lower and *sob sob* rates no longer trundle in without question at 12%, but yr still 'likely' to end up 7-8% RoR ahead...
Perhaps the UK could also approach Canada for political union - her majesty might have to move to Ottawa but it would save us having to go through the costly and unpredictable process of Brexit as we would be able to join the Canadian trade agreement with the EU.
Brilliant.
There's been a campaign for years for WA, OR and BC to merge to form Cascadia. Which has to be the worst proposed name for a country that I've ever heard.
A statistic I find fascinating from accident causation research is that, where accidents do happen due to 'human error' (which should generally be treated as a symptom of the system, not a root cause), 85% of people put in that situation would have reacted the same way and caused the same accident. I do not know what the statistics are for criminal behaviour, but it would surprise me if that general principle did not translate.
The other aspect of Richard's response I wholeheartedly agree with is that success breeds complacency, arrogance and laziness, and reduces our willingness to listen and learn, or even to notice that things are beginning to go off the tracks unless we consciously set things up to question ourselves despite our success.
Human error is rarely what the name says it is. It is sometimes human performance variance, but even more usually environmental variance. I would expand upon Richard's lists of measures by more explicitly breaking down management processes - organisational structure and culture, performance management systems (in particular) and alignment of goals with ethics and safety values are absolutely key to getting behaviours you want and minimizing behaviours you don't.
I agree with most of this, particularly your last sentence.
But - and this is a big but - one of the things we have forgotten as a society (and this has been shown up in the City but not just there) is the importance of character and judgment. Controls and processes and all the rest are important. But in the end you cannot - certainly in banking - set down rules and processes for every simple thing which could happen. You have to rely on people who have the right character and judgment when they have to make a decision when there is no rulebook, when there is no manager to ask, when there is no-one looking.
And what you need at that moment - and there are far more such moments in life than we perhaps think - is someone whose character, whose judgment, whose default instincts, are to try and do the right thing. Hard to define - but you know it when you see it - and when you don't.
Judgment is like a muscle. The more you use it the better judgment you develop. We have as a society I think forgotten its value. That is one reason why we got a lot of highly intelligent people in the City behaving in scummy ways. They forgot to ask themselves the question that we should all - almost without realizing it - be asking ourselves on a daily basis: "Not just can I do this?" (i.e. do the rules permit it). But "Should I do it?"
The difference between "can" and "should" is what a professional, moral conscience - judgment, if you will - is.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
(emphasis added)
I hope you don't risk real money at poker thinking you can beat what you know to be a perfectly-playing opponent.
Only ever play poker with friends for beer money, as an alternative to an expensive night in a bar.
It's one of those games that's simple enough on the face of it, but those who do it professionally are *really* good at it.
It's not really a card game, though, is it? Sure, calculating the probabilities is what all the TV shows do for you. But ultimately it is psychology and micro-expressions.
Channelling trump...... WRRRRRRRRRONG
In what way is it wrong. My point is that you start with the probabilities given by the cards, then overlay that with the probability of the player bluffing in general based on your knowledge of how they play the game and how many times they might sacrifice bluff in order to mislead, and further overlay that with the probability of the particular player bluffing in this particular instance based on tells and 'gut feeling', itself simply a subconscious probability calculator.
Perhaps the UK could also approach Canada for political union - her majesty might have to move to Ottawa but it would save us having to go through the costly and unpredictable process of Brexit as we would be able to join the Canadian trade agreement with the EU.
Brilliant.
There's been a campaign for years for WA, OR and BC to merge to form Cascadia. Which has to be the worst proposed name for a country that I've ever heard.
AI discussion in regards GO and Chess. Looks like we will be adding poker (HU no limit hold'em) to that list. CMU bot is currently beating 4 top pros fairly easily under conditions where they take out all-in variance and mirror hands so the bot plays either side of hands.
Poker should be 'fairly' easy for a computer shouldn't be. Being able to measure and calculate Odds and expected payouts.
The AI would do the maths and probability really well, but it won't beat humans until it learns to bluff. A perfectly-playing machine opponent would be easy for a skilled player to beat. If it bets big, you should always fold.
(emphasis added)
I hope you don't risk real money at poker thinking you can beat what you know to be a perfectly-playing opponent.
Only ever play poker with friends for beer money, as an alternative to an expensive night in a bar.
It's one of those games that's simple enough on the face of it, but those who do it professionally are *really* good at it.
It's not really a card game, though, is it? Sure, calculating the probabilities is what all the TV shows do for you. But ultimately it is psychology and micro-expressions.
Channelling trump...... WRRRRRRRRRONG
In what way is it wrong. My point is that you start with the probabilities given by the cards, then overlay that with the probability of the player bluffing in general based on your knowledge of how they play the game and how many times they might sacrifice bluff in order to mislead, and further overlay that with the probability of the particular player bluffing in this particular instance based on tells and 'gut feeling', itself simply a subconscious probability calculator.
There is a GTO solution to HU no limit poker. It require no use of tells or gut feeling or even adjusting to how your opponents plays.
In fact armed with the GTO solution, you could tell your opponent your exact cards and your strategy given every scenario...and even with all that information they would not be able to beat you.
Humans have coded a program which provides a near perfect estimation of a Nash Equilibrium solution to limit Heads Up Poker. You can actually go on the web and play it, and they will show you its strategy in every situation, and you still can't beat it.
For No Limit, there is still a way to go, but the current bot playing the humans is beating 4 of the top 20 HU specialists in the world, having played trillions of hands against itself in search of a Nash Equilibrium solution.
Perhaps the UK could also approach Canada for political union - her majesty might have to move to Ottawa but it would save us having to go through the costly and unpredictable process of Brexit as we would be able to join the Canadian trade agreement with the EU.
Brilliant.
There's been a campaign for years for WA, OR and BC to merge to form Cascadia. Which has to be the worst proposed name for a country that I've ever heard.
I love that article, so well written. On a cheerier note, it appears that more recent work has all but discredited the alarming forecasts about the collapse of Cumbre Vieja in the Canaries. So at least the East Coast will be safe .
But - and this is a big but - one of the things we have forgotten as a society (and this has been shown up in the City but not just there) is the importance of character and judgment. Controls and processes and all the rest are important. But in the end you cannot - certainly in banking - set down rules and processes for every simple thing which could happen. You have to rely on people who have the right character and judgment when they have to make a decision when there is no rulebook, when there is no manager to ask, when there is no-one looking.
And what you need at that moment - and there are far more such moments in life than we perhaps think - is someone whose character, whose judgment, whose default instincts, are to try and do the right thing. Hard to define - but you know it when you see it - and when you don't. ...
This I very strongly agree with. It's one of the ironies of regulation: the more regulation you apply, and the more detailed it becomes, the more you discourage the type of big-picture moral judgement you describe. You can see why: "Here's a chance to make money, it can't be dodgy because I've checked our 10,000 page book of rules and it's not forbidden". Excessively detailed regulation is itself a moral hazard, a sort of moral infantilisation.
I agree with most of this, particularly your last sentence.
But - and this is a big but - one of the things we have forgotten as a society (and this has been shown up in the City but not just there) is the importance of character and judgment. Controls and processes and all the rest are important. But in the end you cannot - certainly in banking - set down rules and processes for every simple thing which could happen. You have to rely on people who have the right character and judgment when they have to make a decision when there is no rulebook, when there is no manager to ask, when there is no-one looking.
And what you need at that moment - and there are far more such moments in life than we perhaps think - is someone whose character, whose judgment, whose default instincts, are to try and do the right thing. Hard to define - but you know it when you see it - and when you don't.
Judgment is like a muscle. The more you use it the better judgment you develop. We have as a society I think forgotten its value. That is one reason why we got a lot of highly intelligent people in the City behaving in scummy ways. They forgot to ask themselves the question that we should all - almost without realizing it - be asking ourselves on a daily basis: "Not just can I do this?" (i.e. do the rules permit it). But "Should I do it?"
The difference between "can" and "should" is what a professional, moral conscience - judgment, if you will - is.
What you call character and judgment is what organisational culture should be there to provide. I describe organisational culture as what people do when no-one is looking. And it is possible to create and manage a culture so that, when the rulebook does not provide specific or applicable guidance, people will generally make the decision you'd prefer them to.
This is a key principle from Tom Peters classic, In Search of Excellence. By creating a strong culture around a few key values, you can give the workforce far more autonomy because you can trust that decisions made will not betray the core values of the organisation.
HROs cannot work without autonomy as decisions have to be made rapidly in unpredictable changing environments by those with the most knowledge of that particular situation, invariably not the executives both those on the front line. Thus culture becomes essential.
The financial industry is a 'high consequence industry' in that even small errors can cause massive consequences. It has much to learn from HRO principles and practice. If I have not already recommended it to you, try Sutcliffe and Weick's Managing the Unexpected: Resilient Performance in an Age of Uncertainty.
For a more fun read about a very practical application of organizational culture that ultimately saved lives, Shackleton's Way by Morrell and Capparrell.
There is a GTO solution to HU no limit poker. It require no use of tells or gut feeling or even adjusting to how your opponents plays.
In fact armed with the GTO solution, you could tell your opponent your exact cards and your strategy given every scenario...and even with all that information they would not be able to beat you.
Humans have coded a program which provides a near perfect estimation of a Nash Equilibrium solution to limit Heads Up Poker. You can actually go on the web and play it, and they will show you its strategy in every situation, and you still can't beat it.
For No Limit, there is still a way to go, but the current bot playing the humans is beating 4 of the top 20 HU specialists in the world, having played trillions of hands against itself in search of a Nash Equilibrium solution.
I should clarify a couple of points...
1) The GTO solution require absolutely no consideration of your opponent or their strategy. All you need is the game state, from which you will have a mixed strategy response.
2) When I say show them your exact cards, by that I mean before community cards are dealt..as obviously if you could see your opponents cards on the river that would be a different matter.
3) For multi-player games of poker ( > 2), it is still debated if there is a GTO solution to this game, mainly because your opponents could collude against you. However, given the presumption that every other opponent is playing to win for themselves, extremely strong strategies already exist that again take no consideration of opponent strategy (in fact they have been found playing as bots on major sites and winning millions).
But - and this is a big but - one of the things we have forgotten as a society (and this has been shown up in the City but not just there) is the importance of character and judgment. Controls and processes and all the rest are important. But in the end you cannot - certainly in banking - set down rules and processes for every simple thing which could happen. You have to rely on people who have the right character and judgment when they have to make a decision when there is no rulebook, when there is no manager to ask, when there is no-one looking.
And what you need at that moment - and there are far more such moments in life than we perhaps think - is someone whose character, whose judgment, whose default instincts, are to try and do the right thing. Hard to define - but you know it when you see it - and when you don't. ...
This I very strongly agree with. It's one of the ironies of regulation: the more regulation you apply, and the more detailed it becomes, the more you discourage the type of big-picture moral judgement you describe. You can see why: "Here's a chance to make money, it can't be dodgy because I've checked our 10,000 page book of rules and it's not forbidden". Excessively detailed regulation is itself a moral hazard, a sort of moral infantilisation.
Agree. Try getting that message across to regulators, though. Or even to those in banks drafting endless policies.....
There is a GTO solution to HU no limit poker. It require no use of tells or gut feeling or even adjusting to how your opponents plays.
In fact armed with the GTO solution, you could tell your opponent your exact cards and your strategy given every scenario...and even with all that information they would not be able to beat you.
Humans have coded a program which provides a near perfect estimation of a Nash Equilibrium solution to limit Heads Up Poker. You can actually go on the web and play it, and they will show you its strategy in every situation, and you still can't beat it.
For No Limit, there is still a way to go, but the current bot playing the humans is beating 4 of the top 20 HU specialists in the world, having played trillions of hands against itself in search of a Nash Equilibrium solution.
But - and this is a big but - one of the things we have forgotten as a society (and this has been shown up in the City but not just there) is the importance of character and judgment. Controls and processes and all the rest are important. But in the end you cannot - certainly in banking - set down rules and processes for every simple thing which could happen. You have to rely on people who have the right character and judgment when they have to make a decision when there is no rulebook, when there is no manager to ask, when there is no-one looking.
And what you need at that moment - and there are far more such moments in life than we perhaps think - is someone whose character, whose judgment, whose default instincts, are to try and do the right thing. Hard to define - but you know it when you see it - and when you don't. ...
This I very strongly agree with. It's one of the ironies of regulation: the more regulation you apply, and the more detailed it becomes, the more you discourage the type of big-picture moral judgement you describe. You can see why: "Here's a chance to make money, it can't be dodgy because I've checked our 10,000 page book of rules and it's not forbidden". Excessively detailed regulation is itself a moral hazard, a sort of moral infantilisation.
In safety, I rail against regulation. Regulation breeds compliance, which is a minimum effort mindset. Excellence - be it safety, moral behaviour or even production - requires discretionary effort and continual learning. Compliance is not just neutral to discretionary effort and learning, it actively shuts it down by undermining autonomy (and far less importantly detracts resources away from most useful efforts).
Comments
He sympathised with me: "Public transport is horrible, isn't it. I sacked my pilot yesterday for shouting at air traffic control - and it had to be when the chopper was in for a service. Would you believe that I had to fly BA?"
You may want to match the liabilities with pension maturity to lock in the fund at a known cost. Some pension fund managers may be able to do this more cheaply than others. Academic studies have shown the best lon term fund managers are those with the lowest cost (or those with inside information who use it illegally).
Would the UK government step up to bail out Euroclear if it were to go bust leaving those with Sterling contracts out of pocket? Likewise, would the ECB/EU step in if it was Euro denominated deals at the LCH?
It's not though, or at least in depth. Go, like Chess, has some fairly simple rules that are well defined. It's incredibly shallow machine learning. Try defining similar rules to act as bounds for (say) driving: it's massively more complex problem than just inputting the highway code.
The vast majority of AI mentioned in the press is just smoke and mirrors. People are (justifiably) sceptical about media reports on the economy or politics, yet swallow any old guff from a journalist when it comes to technology.
As a related aside, I wonder how the Go versus human games work on different size boards? I wonder if a larger or smaller board makes the games easier or harder for the machine to win.
It is not the level of interest rates which matter but
a) the interest margin between borrowing and lending rates and
b) how much of the orginal loan gets paid back over the loan period.
https://twitter.com/sionsimon/status/822402041716744193
I'm sure this Damascene conversion is entirely principled and nothing to do with electoral calculation.
P.S. I accidentally clicked (and then unclicked) the "Off Topic" instead of the "Quote" button. I hope the system doesn't hold it against you. What purpose does the "Off Topic" button serve here anyway?
Commercial banks are the middle men for debt and investment banks are the middle men for equity.
Jeremy Corbyn: "too simplistic" to see Scottish problems as national questions & north English ones as economic. It is all people vs capital
Ah yes. There's nothing simplistic about Jeremy Corbyn's analysis.
The larger the board, the bigger DeepMind's advantage.
Shockingly, I already have quite a lot of off-topic notices (41). I like to think at least some of them were for winning tips
It's very noticeable how much less input I have to process when driving on a quiet dual carriageway compared to going through the centre of Cambridge in dusk at rush hour in the rain.
This is why driving on a dual carriageway can be so less stressful than driving through a city: you are doing much less work. It's also why they've been able to make good driverless cars on freeways / motorways long before they've got towns and cities sorted.
I'm glad they've moved the writ quickly at any rate.
I'll be more impressed when a program coded to beat someone at chess can beat someone else at go, with no programming except knowing the rules. Oh, and then writ a novel.
Deepmind are doing some interesting stuff, though I'm wary about them getting access to medical records.
https://www.theguardian.com/politics/blog/live/2017/jan/20/politics-live-readers-edition-friday-20-january
It's interesting because it starts logically enough but the logic leads inexorably to a completely nutty conclusion. He correctly in my view puts his finger on a key point (his para 6), which is that any successful attempt by parliament to put restrictions on the PM's negotiating position would inevitably mean we get a less favourable deal from the EU.
After that, he descends into an amusingly absurd fantasy.
It all serves to show that Labour really are in a bind about this. The LibDems perhaps less so, because they are not pretending to be a party of government.
It's more likely that we should be having a 'banks will get disintermdiated' debate. They've protected their business model for so long that they seem to have fialed to spot inherent weaknesses. Weaknesses that fintech is rapidly exploiting. If you want t:
1. Borrow
2. Raise equity
3. Exchange forex
4. Invest
5. Blah blah blah
there are an ever increasing range of options for customers that don't involve a bank at all.
The City should be MUCH more concerned about its business model than its location. And we should all push to make London the absolute centre of European fintech. Oh...it is. :-)
I believe he did write the most linked article in PB's history.
Perhaps someone has a link to it.
Of course CA has always had a soft spot for shysters and mountebanks. Perhaps Nige can found his own religion.
https://thinkpol.ca/2016/11/12/canadians-invite-california-oregon-and-washington-to-join-canada/
Perhaps the UK could also approach Canada for political union - her majesty might have to move to Ottawa but it would save us having to go through the costly and unpredictable process of Brexit as we would be able to join the Canadian trade agreement with the EU.
Probably.
An autonomous F1 car would be very uncompaetitive against real racing drivers, the AI would be way too conservative with things like yellow or blue flags, and would struggle to overtake in situations where the other driver has to be co-operative. In F1 specifically it would also struggle with race strategy and tyre management without a team of humans advising it as happens with a real driver.
A developed-from-scratch autonomous racing car will be very quick on the brakes and around the corners, not having to worry about getting a stiff neck after a couple of hours' racing.
But let's face it, a huge draw of the F1 race is the chance of a big accident, and seeing someone either walk away or end up in the hospital. It's one of the few things left which is popular because it's bloody dangerous - even if the safety record is massively improved in recent times.
https://youtube.com/watch?v=joIsgP9StAY
Bluffing could be programmed in as well.
I hope you don't risk real money at poker thinking you can beat what you know to be a perfectly-playing opponent.
Such a solution does not need to consider the opponent's strategy (or adjust) & we know that the GTO solition is a mixed strategy ie you don't just make a certain move in a particular scenario , rather it is about the proportions you take each possible action ie raise, call, fold when facing a bet.
2-3 years ago a basically near perfect GTO solition for limit HU poker was found. NL is still a long way off, but it seems to approximated GTO solution is looking better than some of the best humans despite being miles away from GTO ie humans were no were near as good as they thought.
One was criminal. The other isn't.
Oh dear! A lot of whataboutery from you. I'm a bit surprised.
Of course bad behavior exists in other organisations: look at MPs for instance and the police. But I was writing about the City and pointing out the gap between its high opinion of itself, the reality and what that meant for public policy at a time when the City is under pressure and could do with some friends.
Its past behavior and willingness to excuse itself on the grounds of its revenues is not a sufficient answer anymore. Not that it should ever have been, IMO. The City is learning the hard way now the opposite of the Boy Crying Wolf lesson i.e. if you spend all your time saying how wonderful you are and then it turns out that you're not wonderful, don't be surprised when people don't listen to you when you say how important you are to them.
For an industry which has a lot of clever people working in it, it is remarkable how few intelligent - and emotionally intelligent - spokespeople it has. I cringe when I hear someone speaking for the City. They invariably get the tone (as well as the content) wrong. May is not listening to the City because she is listening to others. But she is also not listening to the City because the City's case is all too often presented with a mix of bluster and arrogance and "don't you know who I am" which grates on those who do not see it the way the City would like to be seen.
Christ, if it grates on me - and I've more experience than most of it - imagine how it grates on people in Barrow earning a pittance by comparison.
It's one of those games that's simple enough on the face of it, but those who do it professionally are *really* good at it.
Aside from shooting their representative in the face with a Taser, how could that possibly cause problems!
This is very similar to chess. After deep blue, chess players realises they werent anywhere near "perfect". Today's top chess players are miles better because they can see the computer learned stragies and build on them. Apparently computer can beat the beat human at chess, but human with computer assistance beats the computer on its own.
"Sometimes nothing can be a real cool hand"
https://en.wikipedia.org/wiki/Cascadia_(independence_movement)
Fake news?
https://www.youtube.com/watch?v=Fb8JTFxc2Xg
http://www.newyorker.com/magazine/2015/07/20/the-really-big-one
It's really, really scary.
Translating what Richard says into what I do, he is essentially saying that the way organizations have set themselves up, they way they measure themselves (e.g. waiting lists for NHS) and the way that they reward people drives most of the behaviour. That is the essence of systems approaches to safety (and excellence).
Whether the something that is wrong (because it leads to suboptimal outcomes) is criminal or not really is arbitrary based on what we choose to legislate. Even criminal behaviour happens in large part because that is what the system sets up some fairly predictable number of people to do.
A statistic I find fascinating from accident causation research is that, where accidents do happen due to 'human error' (which should generally be treated as a symptom of the system, not a root cause), 85% of people put in that situation would have reacted the same way and caused the same accident. I do not know what the statistics are for criminal behaviour, but it would surprise me if that general principle did not translate.
The other aspect of Richard's response I wholeheartedly agree with is that success breeds complacency, arrogance and laziness, and reduces our willingness to listen and learn, or even to notice that things are beginning to go off the tracks unless we consciously set things up to question ourselves despite our success.
Human error is rarely what the name says it is. It is sometimes human performance variance, but even more usually environmental variance. I would expand upon Richard's lists of measures by more explicitly breaking down management processes - organisational structure and culture, performance management systems (in particular) and alignment of goals with ethics and safety values are absolutely key to getting behaviours you want and minimizing behaviours you don't.
1. The City has cost us a lot of money as well as making us a lot in the past. Money incites emotion and it is not just about envy. Look at peoples' reaction to MPs fiddling their expenses. They didn't get so het up about police misfeasance.
2. Banking is about trust. It's the only thing which matters in banking. That trust has been broken or very severely damaged. And once you lose trust it takes an age to rebuild it. And those who have had their trust abused can feel vitriolic about it. Understandably.
It is far too simplistic (and comforting) to say that it is just envy. If you do you risk IMO missing the importance of the very fundamental failings which recent scandals have shown us and, more alarmingly, you are more likely not to take a good hard look at yourself and really take the steps needed to improve matters. This is hard. And people tend to shy away from hard steps like this. Especially if there are people around to say that the criticisms are all down to envy, blah, blah....
Are the changes that fin tech is already bringing to commercial SME lending such as P2PFA, going to cause radical changes throughout the sector? With 50% growth pa in P2PFA, the oligopolies of main street banks may look time limited. Note that peer to peer now starting into mortgages albeit at very very small levels. Here is one recent article.
http://www.businessinsider.com/p2pfa-q4-2016-stats-strong-end-to-year-p2p-funding-circle-2017-1?r=UK&IR=T
Both myself and Rod formerly of this parish have invested in this one. Less defaults, better security and higher rates than Funding Circle.
The loanbook isn't as mature, availability is lower and *sob sob* rates no longer trundle in without question at 12%, but yr still 'likely' to end up 7-8% RoR ahead...
But - and this is a big but - one of the things we have forgotten as a society (and this has been shown up in the City but not just there) is the importance of character and judgment. Controls and processes and all the rest are important. But in the end you cannot - certainly in banking - set down rules and processes for every simple thing which could happen. You have to rely on people who have the right character and judgment when they have to make a decision when there is no rulebook, when there is no manager to ask, when there is no-one looking.
And what you need at that moment - and there are far more such moments in life than we perhaps think - is someone whose character, whose judgment, whose default instincts, are to try and do the right thing. Hard to define - but you know it when you see it - and when you don't.
Judgment is like a muscle. The more you use it the better judgment you develop. We have as a society I think forgotten its value. That is one reason why we got a lot of highly intelligent people in the City behaving in scummy ways. They forgot to ask themselves the question that we should all - almost without realizing it - be asking ourselves on a daily basis: "Not just can I do this?" (i.e. do the rules permit it). But "Should I do it?"
The difference between "can" and "should" is what a professional, moral conscience - judgment, if you will - is.
In fact armed with the GTO solution, you could tell your opponent your exact cards and your strategy given every scenario...and even with all that information they would not be able to beat you.
Humans have coded a program which provides a near perfect estimation of a Nash Equilibrium solution to limit Heads Up Poker. You can actually go on the web and play it, and they will show you its strategy in every situation, and you still can't beat it.
For No Limit, there is still a way to go, but the current bot playing the humans is beating 4 of the top 20 HU specialists in the world, having played trillions of hands against itself in search of a Nash Equilibrium solution.
This is a key principle from Tom Peters classic, In Search of Excellence. By creating a strong culture around a few key values, you can give the workforce far more autonomy because you can trust that decisions made will not betray the core values of the organisation.
HROs cannot work without autonomy as decisions have to be made rapidly in unpredictable changing environments by those with the most knowledge of that particular situation, invariably not the executives both those on the front line. Thus culture becomes essential.
The financial industry is a 'high consequence industry' in that even small errors can cause massive consequences. It has much to learn from HRO principles and practice. If I have not already recommended it to you, try Sutcliffe and Weick's Managing the Unexpected: Resilient Performance in an Age of Uncertainty.
For a more fun read about a very practical application of organizational culture that ultimately saved lives, Shackleton's Way by Morrell and Capparrell.
1) The GTO solution require absolutely no consideration of your opponent or their strategy. All you need is the game state, from which you will have a mixed strategy response.
2) When I say show them your exact cards, by that I mean before community cards are dealt..as obviously if you could see your opponents cards on the river that would be a different matter.
3) For multi-player games of poker ( > 2), it is still debated if there is a GTO solution to this game, mainly because your opponents could collude against you. However, given the presumption that every other opponent is playing to win for themselves, extremely strong strategies already exist that again take no consideration of opponent strategy (in fact they have been found playing as bots on major sites and winning millions).