slowjoe

I have a confession to make. I've been following the "History of Rome" podcast series by Mike Duncan. More interesting to me than Hannibal's victories, or indeed the the Battle of the Teutoberg forest, is hints at the change of the mood of the times.

A republic built on regular elections devolved through Proscription lists and regular civil wars, to the point that no-one wanted to contest the power of the Augustus and his successors, for fear of further blood-letting. At the time of the defeat of Hannibal by Scipio, could this have been forseen, and perhaps stopped? Gifted with precognition, could changes have been made before the Gracchi, Marius, Sulla and the two sets of triumvirs been forestalled? Or would a man blessed even with such precognition plus the abilities and lack of ambition of an Agrippa be doomed merely to relive the frustrations of Cassandra?

An amusing exercise.

I likewise wonder if we may be moving along a similar course in the western world. There are men and women (James Hansen and George Monbiot spring to mind) who believe in climate-crime trials for oil and coal company executives. I think there is a parallel between this and proscription lists. I likewise see the throwaway comment about "tempted to beat the crap out of" a climate sceptic named Pat Michaels by Dr. Ben Santer along with group-think and subverting the peer review process as less than reassuring. Similarly, one of my UK friends talks about treason trials for the current Labour government, and has a number of charges to choose from. Of course, the first trial of this kind is merely the first trial of many. Were Hansen to succeed in establishing his climate court, I'd be surprised if none of his allies were to fall under scrutiny of it, or something similar.

I write this shortly after watching the video of an MIT forum on Climategate. One comment by Dr. Judith Layzer about the state of the American media hit home with me. She said that there was no media as there used to be (meaning a neutral honest broker reporting both sides of an issue). There were polarised echo chambers without dissenting views. Assuming positive reinforcement, then surely this is recipe to create a political powderkeg?

How to diffuse the charge? Well, we now have significant elements of the 19th and 20th century civil fabric facing serious challenge. If the media is polarised, then compromise tends to be vilified, and the appeal to the median voter has less appeal. I suspect that this problem will resolve over the next 20 years as the old press loses power and influence in the way of the canal system has. Of course, we have to survive those 20 years first.

We have similar troubles with science and universities. The Climategate emails and documents and Hockey stick controversy show that the peer review process has problems. The fact is that the peer review works at the pace of the pre-internet world. It isn't robust against the incentives structure of "publish or perish", Black Swan style non-linear rewards, let alone the pressures of politicised science. Tenure committees stacked with one political viewpoint (whether right or left) is unhealthy. It is ironic that the left today has a belief in diversity which omits protection for political views.

If the university world turns to the liberal, then the military world has tended to go to the right. That too isn't healthy. If institutional group-think is a problem in universities or the media, how much more dangerous could it be among the rough men charged with protecting us? How far from a citizen-army reflecting the diversity of the citizenry is it healthy to go?

So, the question is, what small number of civil reforms would you advocate to improve the level of civility nad the effectiveness of civil society?

I've written about peer review a little while ago here and here.

One of the better meta-analyses of ClimateGate has been written by Megan McArdle. The fact that she quotes one of the better pieces of writing by Richard Feynman gets her extra points :)

The piece she quotes describes a multi year process of drift of the published charge of the electron following the famous Millikan Oil Drop experiment. Feynman argues that the results were biased by the experimenters expectations. McArdle then says that the same kind of "expectation bias" might be in play:


That is the actual worrying question about CRU, and GISS, and the other scientists working on paleoclimate reconstruction: that they may all be calibrating their findings to each other. That when you get a number that looks like CRU, you don't look so hard to figure out whether it's incorrect as you do when you get a number that doesn't look like CRU--and maybe you adjust the numbers you have to look more like the other "known" datasets. There is always a way to find what you're expecting to find if you look hard enough.


To me, that's a natural process. I'd expect people to be doing back-of-the-envelope calculations on any results. Scrutiny of "unusual" results is likely to be higher. So it should be.

The following paragraph says:


There are other issues: selection bias in the grant process, papers with large results being much more likely to be published than papers with equivocal results, professors preferring students who agree with them, and so forth. I doubt that could amount to faking the entire thing. But it could amplify the magnitude.


Most of this is less acceptable. The quotes in the CRU emails which talk about sacking editors, and blocking papers from the IPCC are dishonest (but not crinimal). And a significant problem comes from not understanding the implications of the statistical tools used. For instance, the original hockey stick comes originally from a modified (ie incorrectly calibrated) Primary Component analysis + misuse of Bristlecones as a temperature proxy. After the mistake has come 10 years of defence. It is this defence that is causing many of the problems.

It is well to remember what I call the "Fuck-up Theory of History". Anywhere that someone claims there is a conspiracy, check if you can blame incompetence instead. The idea of a vast conscious climate conspiracy is unlikely beyond believe. Apart from anything else, the Climate campaigners are accumulated at the liberal end of the political spectrum. This area is full of people who assign weight to hope, aspiration and intention compared to the right wing. They could be duped, bias themselves, but not engage in a vast conspiracy.

The TV segments I've seen show this. The defence talking point is to counter-attack a claim of conspiracy. They are right that there isn't sufficient evidence for this claim, but generally, this isn't the claim which is being made.

Climategate is causing a lot of cognitive dissonance at the moment. People who might call themselves "audit-geeks", who have bookmarks for Bishop Hill, Steve McIntyre or Anthony Watts are in discussions with people who are going "trick? FOI? Um how does this fit in?" There are a couple of problems here. Managing and organising the information which is passed on is tricky. There is a lot of detail.

Secondly, there is a trap of making statements which is excessively strong based on the facts as known. What we have is a situation where there is prima facie evidence of fraud and conspiracy. There isn't proof, however. The correct response is to audit all the science, and stall on government action. But the basis for this is a Bowdlerised version of the Hyppocratic Oath: first, do no harm.

The temptation is to repeat the crime of the CRU cabal and overstate the evidence.

I did a brain dump over two posts this weekend in a comment at Andrew Prock's Livejournal. Andrew is comfortably within the AGW camp, and as such, I mis-characterised his post as fightback. Instead, he rather acutely pointed out that the whole debate is becoming partisan and rhetoric-driven. (Update: see comment for correction.) Anyway, the following was my brain-dump.

Post 1


That's almost a thesis question. But I know a little bit about this, so I'll give you my understanding.

In the beginning, there were two questions asked about global warming.
1. Is this warming unprecedented? Didn't we have the Medieval Warm Period 1000-1400 and the Little Ice Age 1600-1750?

2. How much of the warming that we observe comes from the Urban Heat Island effect?

Point 1: Unprecedented warming

The hockey stick came from a paper now referred to as "Mann, Bradley, Hughes 98" or "MBH98". This paper used different "temperature proxies" (typically ice cores and tree rings) to reconstruct temperatures going back to about 1000AD. In order to do this, each proxy had its average subtracted from it (to base it at zero), and then divided by its standard deviation. This may also have been the paper where the proxy was spliced with the temperature record (this being refered to now as "Mike's Nature Trick") The resulting data was fed into a hacked statistical algorithm called something like Primary Component Analysis (PCA). This paper said that the recent warming was unprecedented, but also that the LIA and MWP were not much different from the millenial average.

A retired statistician named Stephen McIntyre and an academic named Ross McKitrick tried to reproduce this. Mann refused to provide either data or code. Nevertheless, M&M found that when they re-wrote Mann's PCA, they generated Hockey sticks from red noise. In addition, they found that from the list of proxies in the paper, that one group, from "strip-bark bristlecone pines" in the western US seemed to provide all of the warming signal.

M&M had major problems publishing these findings. These problems are possibly related to the stuff about not recognising journals, pressure to remove journal editors, and pressure on academic departments to stop retired staff using the university name. The original journal wanted to keep any comment to a small word limit. When they rewrote as a paper, they were told there wasn't sufficient material.

Mann refused to provide his data or code, but this material was found a couple of years ago on an FTP server.

Anyway, while M&M were reverse engineering the MBH98 paper, other reconstructions were being written. Among the proxies they use, there were the bristlecones plus either a dataset called "Polar Urals" or, later, "Yamal", both in Siberia, and published by the deputy head of CRU, Keith Briffa. (Briffa was also the editor of the historical temp chapter in the latest IPCC report.) Polar Urals had a problem: it said that a year between 1000 and 1100 (bang within the MWP) was the coldest in the past millenium. Yamal didn't have this problem, so later reconstructions have used this.

The Yamal data was subject to a bunch of FOI requests and journal data access requests. Finally, it was used by the original author last year in a paper to Philosophical Transactions of the Royal Society B. The Royal Society has a policy of access to data and code for papers published in its journals (similar to other journals where previous papers have been published.) However, the Royal Society enforced this. This fall, the Yamal data was made available. For the period after 1990, less than 10 trees were used, and one tree in particular was 6 standard deviations away from the average.

All papers which produce a hockey stick use at least one of Polar Urals, the Greybill Bristlecones or Yamal. In response to the critics, papers were written that showed if you had use B and Y, then removed either, you retained a hockey stick.

The latest papers also use a lake sediment dataset from a lake in Finland as a temperature proxy. However, they have inverted the series from the interpretation in the original academic paper, and refused to address the point. This series was used upside down in a paper that Mann released this week.

That is as much as I know about point 1 above.


Post 2


Point 2: Urban Heat Island
It's actually hard to produce a global temperature. You can look at the satellite data, the land data or the sea data. The satellite data shows less warming than the land, which has been explained as orbital anomalies, and an adjustment is added to the satellites.

There are a few indexes, among which is HADCRUT compiled by CRU (of leak fame), and another compiled by NASA's Goddard Institute of Space Science (headed by James Hansen). HADCRUT uses land stations. One of the problems with land stations is that the perfect weather station should be rural, unshaded, well maintained and in the centre of about an acre of ploughed earth. They are supposed to be in a whitewashed wooden box a standard height above ground with a north facing door. Very few are like this. Likewise, the ideal station should not have moved.

In contrast, the typical weather station may or may not be shaded, often isn't in the same location where it started, almost certainly isn't in the centre of that ploughed earth, and may be in the centre of a town, or at least near other man made structures. There are examples of stations with south facing doors, electric light bulbs wired into the box for reading (banned), boxes with the wrong paint, placed next to aircon outflows, beside tennis courts or (EXTREMELY frequently) airport runways.

Because of urbanisation and station moves, adjustments are made to the historical data. These adjustments are very controversial, not least because the adjustments have changed in the past 10 years, particularly to make the 1930s colder and the post 1970 period increasingly warmer.

Questions about this in general are not being answered either in the UK or the US.

The FOI requests to the man at the centre of this (Jones) were trying to find out about this, because his academic focus has been the HADCRUT index. However, he stonewalled completely, as described in the article at http://camirror.wordpress.com/2009/11/25/willis-eschenbachs-foi-request/ to the point of refusing to list the stations he used.

In all the above, the reason the data is being requested is that algorithms are not being published and therefore a reverse-engineering exercise is necessary. However, the group around Mann, Jones, Briffa, Schmidt etc are even refusing to provide the input data to validate reverse engineering efforts.

Happy to answer any questions. Some of this may be in error, for which I apologise.


As you can see, there is a body of work which points out significant weaknesses in the work of the global warming science case.

Update:

Andrew corrects me in the first comment below to say that he is in the "trust settled science" camp. I'm happy to correct this mischaracterisation.

Do you have a favourite corner of the web that ISN'T a weblog? Every six months, I go look at a couple of places. Paul Graham is good, and Malcolm Gladwell is also thought provoking.

It is to Gladwell that I go to introduce this piece. A piece on the Enron collapse named Open Secrets claims that the problem of Enron was hidden in open sight, and goes on to list a number of different people who had become bearish on the company well before its collapse.

The piece closes talking about a group of business school students doing a project on the company:


The people in the group reviewed Enron's accounting practices as best they could. They analyzed each of Enron's businesses, in succession. They used statistical tools, designed to find telltale patterns in the company's financial performance — the Beneish model, the Lev and Thiagarajan indicators, the Edwards-Bell-Ohlsen analysis— and made their way through pages and pages of footnotes. "We really had a lot of questions about what was going on with their business model," Krueger said. The students' conclusions were straightforward. Enron was pursuing a far riskier strategy than its competitors. There were clear signs that "Enron may be manipulating its earnings."


And that was two years before the collapse of the company.

There is a question provoked by the article. Is it a fraud or a scandal if the information is out there? Enron provided a pretty full accounting of its SPEs. Thousands and thousands of pages of a details turned up in its annual report. Fine, everyone had read them, right? The BBC and CNN and FOX and Guardian and NYTimes reporters kept us all fully briefed? Oh, they didn't?

I had that "insider" experience myself, in the case of Lehman's Brothers. I had picked up on rumours that Lehman's had problems, and one morning in May 2008, I found myself looking at the cost of November put options on the Lehman's stock. I can place the data simply by looking at the historic stock chart. In the end, I had no idea how to buy options on US stocks, so I didn't go through with it. However, the trade I had planned would have paid off at 1000:1. And I'm not a sophisticated investor. I hadn't seen anyone even mention Lehman options. I had just stumbled over information. So Lehmans' collapse wasn't a "shock". It was a slow motion car crash.

The same thing can be argued regarding the so-called "Climategate" affair. I've been reading Stephen McIntyre's blog since 2004 or so, not as a daily read, but somewhere I go to once a quarter or so. I knew in 2005 that Phil Jones wasn't generous with his data. I knew that he'd changed his story to "the data is lost" in August 2009. That made the headlines in the UK this last weekend, 2009-11-29.

So the whole Climategate thing has been kind of surreal for me. I've known there was a mind-set (conspiracy?) within both CRU and their acquaintances in other institutions not to share data. For instance, there's an email where a climate scientist says something like "I'm not giving the stuff you've asked for, and I've cc-ed all my co-authors on this, so they won't either". So the FOI emails are a surprise? PLEASE!

The sleight of hand which gave us Hockey Stick graphs where tree data cut over to temperature data (known in the trade as "the Divergence Problem": the tree data and thermometer data diverges...) is revealed, and the world is shocked, shocked that this is happening? PLEASE!

But the crowning glory of the CRU/GISS/Mann group is that they bear a more than passing resemblance to the Wizard of Oz. Remember him?

Not the special effects, although there is that.

Not the "running the country". They didn't quite manage that. And they were aiming at the whole world economy.

No, McIntyre and friends, cast as Toto, pulled back the curtain several years ago. And the scientists been saing "pay no attention to the man behind the curtain" for years.

AND IT WORKED!

Via Patrissimo, I found a fitness program which I'm dabbling (gently) in Crossfit. (If you were actually inspired to try Crossfit, the warmup would be where I would start. It took me 2-3 months to be able to do the warmup.)

Anyway, so, web-based fitness program yadda yadda pretty cool yadda yadda fat 40 year old yadda yadda gone from 0 to 20 pullups yadda yadda 59 seconds.

What, you didn't quite catch all of that? Well, the last bit is, I've decided I want to run a lap of a running track in 59 seconds. Why the arbitrary target? Well, one of the Crossfit gyms has put this pdf online. They list different fitness tests, and 4 levels of achievement for each one.

I've managed a couple of "level 3" performances (back squat, row 500m, run 400m), and I'm close in a few others (5kg short on the deadlift, ~3kg short on the military press, power clean vs squat clean on the clean benchmark). Great. But the big game I'm hunting is a level 4.

And lame calf on the list seems to be the 400m run (level 4 is 64 seconds) and my PB (from last year) is 70 seconds. And while I'm hunting big game, I just happened to see that the closest thing to a 4 minute mile I'll ever see is a 59 second 400m.

After all, 6 (or 10!) seconds is only a 10% (16%) speedup, right? It's not so simple. This has been on my mind since last year. I pulled my quad a couple of times around Easter on the track. And my 200m time is currently 30s. I've got some distance to go. But I think I'm going to start updating here with slightly greater frequency, to track my own progress.

Howdy,

I'm playing a little bit of 0.50/1 NLHE on PokerStars. From time to time, I wonder if I'm leaving money on the table. For instance:

PokerStars Game #24000857198: Hold'em No Limit ($0.50/$1.00) - 2009/01/17 16:17:43 ET
Table 'Mentor IV' 9-max Seat #4 is the button
Seat 1: Desti10 ($22.50 in chips)
Seat 2: amazinmets73 ($54.60 in chips)
Seat 3: leroyownsyou ($100 in chips)
Seat 4: Dak9885 ($20.60 in chips)
Seat 5: villian ($108.95 in chips)
Seat 6: hero ($76.60 in chips)
Seat 7: zackpot420 ($62.55 in chips)
Seat 8: bambam500599 ($19.50 in chips)
Seat 9: wjmitchell ($95 in chips)
villian: posts small blind $0.50
hero: posts big blind $1
*** HOLE CARDS ***
Dealt to hero [2d 2c]
zackpot420: folds
bambam500599: folds
wjmitchell: folds
Desti10: folds
amazinmets73: folds
leroyownsyou: folds
Dak9885: folds
villian: raises $3 to $4
hero: calls $3
*** FLOP *** [Qc 2s 6d]
villian: checks
hero: bets $4
villian: calls $4
*** TURN *** [Qc 2s 6d] [5c]
villian: checks
hero: bets $10
villian: calls $10
*** RIVER *** [Qc 2s 6d 5c] [5s]
villian: bets $32
hero: calls $32
*** SHOW DOWN ***
villian: shows [5d 4d] (three of a kind, Fives)
hero: shows [2d 2c] (a full house, Deuces full of Fives)
hero collected $97 from pot
*** SUMMARY ***
Total pot $100 | Rake $3
Board [Qc 2s 6d 5c 5s]
Seat 1: Desti10 folded before Flop (didn't bet)
Seat 2: amazinmets73 folded before Flop (didn't bet)
Seat 3: leroyownsyou folded before Flop (didn't bet)
Seat 4: Dak9885 (button) folded before Flop (didn't bet)
Seat 5: villian (small blind) showed [5d 4d] and lost with three of a kind, Fives
Seat 6: hero (big blind) showed [2d 2c] and won ($97) with a full house, Deuces full of Fives
Seat 7: zackpot420 folded before Flop (didn't bet)
Seat 8: bambam500599 folded before Flop (didn't bet)
Seat 9: wjmitchell folded before Flop (didn't bet)

Should I have lumped in the remaining $25 or so with a raise on the river?

So, I was in the gym (private club in the financial district of London), and for the first time, I saw a client under the age of 18. Well under.

What actually happened was that as I walked out of the men's changing rooms, I passed a 3 year old girl on the way in (trailing a few yards behind her father). I asked a member of staff to have a word.

About an hour later, I returned to the same men's changing room, and the same little girl was singing "I'm in the men's locker room".

What would Jesus you do?

To a freebie reception by DoylesRoom.com via a couple of quick drinks with an exhausted Birks.

Recovering from inaccurate addresses (Shaftsbury Avenue is not Leicester Square), I slowly opened my eyes to how much money is available to online poker dot com marketing.

We had two decent rooms, at least 4 hours of a free bar (pretty much all I can handle on a school night), and ~15 "model | actress | whatever"s to decorate the room.

I started collecting autographs following Birks' hasty departure. I now have a signed copy of the following:

According to Doyle
Super/System
Super/System 2
My 50 Most Memorable Hands

I also got the Mathematics of Poker signed by:

Doyle Brunson
Todd Brunson
Christopher Ferguson

And I need to work in 6 hours.

Finally met one of the London Poker bloggers previously discussed here for lunch. I'm not to kind of bloke to use "delightful company", but if I were, I suspect Mr. Birks would merit that description.

Peter is bubbling with enthusiasm in person (I had anticipated a personality/probability range, like a hand range, which included about 20% probability of someone rather quiet, and that skewed the average.)

Birks is a little like me. We're both playing poker for stakes that pretty much only hurt our pride. On the other hand, I know that a losing session significantly dampens my mood for some time (I'll stop playing for months at a time.)

Why is that? Successful, intelligent middle-aged men worrying about a couple of hundred bucks (which isn't even worth that much in pounds, these days.)

It was pretty much the first time I've had a decent chat with anyone who plays better than me, and I hope to repeat the experience. The catalyst for meeting him was the hand-over of my copy of the Full-Tilt tourney book after a cash/points trade. This leavened the conversation early, since he had browsed it first. I've always been a fan of learning through discussion, so it was very enjoyable.

One thing I noticed about Peter was that he has more sophisticated models of player behaviour than I do, but that they may be a tad rigid. For example, he starts with an assumption that all short-stack players are tight-weak. On the other hand, I typically only think in terms of pot-odds offered against a particular opponent. There is no doubt that talking to him today is worth money in the bank for me, if only from explaining my recent results.

Executive summary:

My employer held a casino-themed Christmas party last week. And I blew a trip to Vegas
on lack of concentration, lack of attention to detail, and believing the boast of a drunk.
But it's ok, cos I got a cool problem to think about.

Longer version:

It was a classic play-money casino, everyone got 100 chips to start with. The games available
were Blackjack, Casino Stud Poker, Roulette and Craps, and there was an undisclosed prize for
the person with the most chips at the end.

I ran my 100 up to about 500 playing Blackjack early on, then switched (after a couple more
beers) to Craps. In between, someone pointed me at the Roulette table, where the big winners
were supposed to be. Someone was supposed to be on 80k already, so as I arrived at the Craps
table, I thought the competition had already And I quickly realised that bet payoffs were wrong.

While the "odds" bets on the pass line were paying off at even money (under-paying), the place
bets were over-paying. I didn't ask what the payout odds were, since I didn't want to tip
the croupier off that he was making a mistake, but he sure as hell wasn't paying off at 7/6 or
7/5.

About halfway through my session, he said that he was going to pay off at 2/1 across the place
bets. Even here, I didn't start increasing my stakes nearly rapidly enough to be optimal. Nor
did I have any kind of accurate count of how many chips I had, or even an accurate idea of the
values of the higher value chips.

Nevertheless, around the time that I was told the table was about to close, I believe I had close
to 10k. Which, according to the best info available to me, wasn't even close to winning. So
I split the stack into 4 piles, and placed it on 5,6,8 and 9. And lost the lot.

In the end, one of my colleagues won 2 flights to Vegas with a chip count of 4990.

Lesson Learned:
1. Play money situations should be taken seriously.

2. What SHOULD my staking pattern have been?

Well, let's look at some situations when the place bets were paying off at 2/1.

Plan 1:
Place 6 and 8.
Payoff  (aka B): 1/1
P(win)  (aka p): 10/36
P(loss) (aka q): 6/36
Optimal stake(from Kelly) = (B*p - q)/B
                          = (1*10/36 - 6/36)/1
			  = 4/36
			  = 1/9

Plan 2:
Place 5,6,8,9
Payoff  (aka B): 1/2
P(win)  (aka p): 18/36
P(loss) (aka q): 6/36
Optimal stake    = (B*p - q) / B
                 = (1/2 * 18/36 - 6/36)/ 0.5
		 = (3/36 *2)
		 = 1/6

Plan 3:
Place 4,5,6,8,9,10
Payoff  (aka B): 1/3
P(win)  (aka p): 26/36
P(loss) (aka q): 6/36
optimal stake   = (b*p - q) / b
                = (1/3 * 26/36 - 6/36) / 1/3
		= (26/96 - 18/96) * 3
		= (8/36) = 2/9

What i don't yet know is which of the above plans is optimal. That's what (both) my readers are for :)

Andy Ward pointed me at Roy Houghton's new poker club, Loose Cannon. It helped that it is directly under my desk at work. (Oh, ok. it might be 30 yards south, as well as down. And the office moving in less than a month. Cut me some literary slack, ok?)

Well, it opened its doors on Monday.

The score is, they are trying to operate as a private members club. Guests will be charged £10 on the door. As an introductory offer, I've got 3 months membership for £18, and I've been told that I'll be charged £5 per visit for that time. However, on Monday, I wasn't charged on the door, so that may be subject to change. In return, they are not charging for any games. There are some backgammon boards, and there is talk of "Sit and Go"s being laid on (there was one SnG on Monday, but the organisation needs tuning...no boards, no game size etc.). The main events are likely to be daily (or twice-daily) poker tournaments.

As I mentioned above, they held a £50 tourney on Monday, kicking off at around 20.30. Starting stacks were T3500. Blind levels were:

25 - 50
50 - 100
75 - 150
100 - 200
150 - 300
250 - 500 (I think, there might have been a 200-400 level)
350 - 700
500 - 1000
700 - 1500 (strange change...)
1000 - 2000
1500 - 3000
2000 - 4000
3000 - 6000
4000 - 8000
6000 -12000 ...
Which is where I got knocked out, in fourth.

Better than a poke in the eye from a sharp stick, I guess.

The final 9 got paid.
The payouts were marginally steeper than Pokerstars:
1st: 33%-ish
2nd: 22%-ish
3rd: 15%
4th: 10%

Room is no smoking, not air-conditioned (yet).

I'll gladly answer any questions my numerous readers (both of you) have.

Scottro is bonus-whoring casinos. I'm following in his footsteps.

So far, one down, roughly 3 hours play, +$124.50.

To start with, I was learning the blackjack strategy and the casino interface, and made some mistakes at 3 x $1 per hand. Later, I doubled to playing 3 x $2 per hand.

It's like stealing candy from a baby.

Over the last 2 weeks (pretty much dating from the completion of HOH III), I've started playing SitnGos again. I'm playing 15+1 Turbos at PokerStars, and I'm up $300 or so over 28 tourneys, for an average of just over $10 per entry.

The stats are rather nice to look at:

Average: $10.64
Hourly rate (ignoring the fact that I'm two-tabling): $20.57/hour
ROI: 62.07%
Std Deviation: $30.98

Wait. What was that last one? Yeah, the standard deviation is $30+ in a $16 tourney. Doubling or quadrupling or 8-plying up the data set reduces the std-dev to $30.71, $30.57 and $30.51 respectively. In other words, the 95% confidence interval includes losing $50 on a $16 buy-in tournament. What this means is that I can't ever use this particular statistical measure to give myself confidence in the fact that I'm a winning player. This is a bit distressing, since all the other stats are what might be called "flattering" stats.

A little digging shows that the confidence interval stuff only applies to normally distributed outcomes, and this is essentially a 4-value outcome dataset. Ok, so can I ask the audience if there are any tools one can apply to SitnGo result sets? Any useful digests that can be applied?

How many tournaments does one need to play before the "short run" morphs into the "medium run" or "long run" for SitnGos?

I've blogged before about getting HOH III a couple of weeks ago.

Harrington has an analysis of Single Table Tournament bubbles in problems 37-42, where he looks at calling an all-in 4-handed in the standard online 50:30:20 payount STT structure we know and love.

He puts the beef in at problem 39, where he works out the equity changes for calling vs folding.

He misses two open goals. First, he doesn't coin the phrase "equity odds" in the following sentence:

Let's make one last observation before we let this 
problem go.  Let's take a look at the risk reward 
ration for Player A, the big stack.  When Player C
call and wins, Player A is the big loser, with an 
equity loss of $101.56.  But when Player C calls and 
loses, most of C's equity goes elsewhere!  
...
Player A's gain is a measly $50.35.  Player A's 
gain-to-loss ratio is almost exactly 2-to-1, meaning
Player A needs a very strong hand as well to justify 
an all-in move.

Why-oh-why didn't they use "equity odds", to complete the
set (with pot odds, and implied odds)?

The other thing they've gotten wrong is that it appears to me that the equity calculations in problem 39 are all bogus. I've made up a spreadsheet for the equities, and I get different numbers to Harrington. My numbers agree with the numbers coming out of the calculator at this calculator.

There is no mention of this on thw two-plus-two forum thread that deals with HOH III errata.

Over the years, I've spent a ridiculous amount of money on Poker learning resources. Owned Syper System over 10 years ago, and I bonus-whored at Doyle's Room to get SS2 (it cost me $135 in bonus.)

I bought Harrington I from High Stakes in Gt. Ormond St. the week it arrived in the UK (and it might have been the day...)

Yesterday, I finally got my hands on Harrington III and The Sklansky and Miller No-Limit books. Two weeks after they were shipped in the states. Conjelco wanted $30+ bucks to ship them to the UK. I have ordered from them direct in the 90s (the Suzuki tourney book, terrible...), and they charged significantly less.

I don't like Conjelco as much any more, and I'm going to let people know it.

I'm currently living in East London (border between Wanstead and Leytonstone for the initiated, or those looking at Streetmap. I've been here since October.

So imagine my surprise when a snooker/poker hall opens 10 minutes from home on foot. They're clearly trying to do a Gutshot, so they've got a bit of an internet cafe thing going where there is PartyPoker, PokerStars, Betfair etc on the desktops of the PCs.

Anyway, there's a £30 tourney this Saturday at 8pm. All entries supposedly need to join the club, so it remains to be seen what time is kick-off. The tourney is a rebuy tourney with a £12,000 guaranteed prize fund. The owners are London greeks, and this shows a little: for example, they refer to "buy-ins" when they mean "rebuys".

Look at there website for details.

Warning: non-poker related material ahead.

Noted British Blogger Iain Dale has published a book authored by a cast of thousands. Together, over the course of a week, the british blogosphere wrote a book cataloguing the crimes of Za Nu Labour party.

For non-Brits, this includes things like accepting a £1 million donation from the head of Formula One motor racing, then interpreting a manifesto ban on "tobacco advertising in sport" to mean "all sport except Formula One2.

Honestly, you have to buy the thing, because by next week, eveyone else will have.

Amazon are cheaper, but Politicos are Iain Dale's publishing company.

You choose.

It has been a strange week.

I had a delivery of a laptop this week. Then I had the same laptop delivered a second time on Thursday.

Should I point this out to the company in question?

I also had an email from Pokerstars. Someone I don't recognise transferred $300 to my account.

Should I point this out?

All views gratefully received.

So, I placed in a multi-table tourney for the first time tonight. I've won places in satellites previously, but never cashed in a tourney. (In fairness, I only played my first MTT of 2005 at the weekend, which obviously had something to do with it...)

For the record, I placed third in an $8 + 0.80 tourney in Doyle's room, getting $239 for my troubles. It had to be done. My gf scored $55 in a free-roll earlier in the week, and is suddenly pushing me to play as well as she does :)

...I think I prefer spending equity to saving it.