Aufheben, Fortunati and Federici

I read Aufheben's review of the Arcane of Reproduction (by Fortunati) years ago. Even though I thought it was a generally correct analysis of a terrible book (or a terrible translation), I thought Aufheben's conclusions were false. I recently read Federici's latest book, Revolution at Point Zero: Housework, Reproduction, and Feminist Struggle. I think I can now effectively say why Aufheben are incorrect.

Aufheben's conclusion about value and housework is:

So then, does housework create value, or not? We have seen in the previous sections that the answer is: no. Housework does not produce commodities, and the labour involved in it cannot be abstracted and measured as abstract labour, as a contribution to value. But we have also seen the value supposedly created by housework cannot be pinned down anywhere.
[Aufheben, The arcane of reproductive production]

I agree that without commodities at some point in the production process, it's impossible to have value. Nevertheless, housework definitely produces and reproduces labour-power. Labour-power can be sold as a commodity. The fact that Aufheben disputes this is ridiculous. Every new generation of worker for the factory, office or farm is created and maintained by house-work. What the hell do they think pregnancy, childbirth, feeding, clothing, caring for, teaching is if it isn't the production of labour-power? What is cooking, ironing, cleaning, washing, sex, etc., if it isn't the reproduction of existing labour-power? This work remains largely un-waged, mostly done by women. Is it relevant that it isn't immediately realised as a wage for it to contain value? It isn't relevant if you accept Tronti's idea of the social factory, as expanded by Federici:

Work appears as just one compartment of our lives, taking place only in certain times and spaces. The time we consume in the “social factory,” preparing ourselves for work or going to work, restoring our “muscles, nerves, bones and brains” with quick snacks, quick sex, movies, all this appears as leisure, free time, individual choice.
[Federici, Silvia (2012-09-01). Revolution at Point Zero: Housework, Reproduction, and Feminist Struggle (Common Notions) (pp. 35-36). Independent Publishers Group. Kindle Edition.]

One way to think of house-work is to understand that the value of house-work is bound-up in the waged-worker and realised when they receive their pay. It's not some sort of secret/hidden extra thing like Fortunati would have you believe, but it's a realisation that people aren't atomic individuals that can be separated out like little units (like bourgeoisie ideology would have us believe).

On the topic of commodity production and who produces them, Marx is clear that it doesn't matter who does it or how it is done once the world-market has been established.

No matter whether commodities are the output of production based on slavery, of peasants (Chinese, Indian ryots), of communes (Dutch East Indies), of state enterprise (such as existed in former epochs of Russian history on the basis of serfdom) or of half-savage hunting tribes, etc. — as commodities and money they come face to face with the money and commodities in which the industrial capital presents itself and enter as much into its circuit as into that of the surplus-value borne in the commodity-capital, provided the surplus-value is spent as revenue; hence they enter in both branches of circulation of commodity-capital. The character of the process of production from which they originate is immaterial. They function as commodities in the market, and as commodities they enter into the circuit of industrial capital as well as into the circulation of the surplus-value incorporated in it. It is therefore the universal character of the origin of the commodities, the existence of the market as world-market, which distinguishes the process of circulation of industrial capital.
[Marx, Capital Volume 2, Chapter 4, "The Three Formulas of the Circuit"]

[...] a commodity produced by a capitalist does not differ in any way from that produced by an independent labourer or by communities of working-people or by slaves.
[Marx, Capital Volume 2, Chapter 19, "Former Presentations of the Subject"]

Once the world-market exists, pretty much everything becomes subject to its rules. A tribe of savages could collectively work together to produce a commodity. Why is a marriage not treated in the same way? The nuclear family expends the labour-power that is realised as exchange-value in the form of the waged-worker's pay cheque. It's a simple as that.

The value of housework can be most clearly revealed through contemporary history. This is because housework is moving from being entirely hidden through the naturalised forms of love and marriage to the waged form.

As the participation of women in waged work has immensely increased, especially in the North, large quotas of housework have been taken out of the home and reorganized on a market basis through the virtual boom of the service industry, which now constitutes the dominant economic sector from the viewpoint of wage employment. This means that more meals are now eaten out of the home, more clothes are washed in laundromats or by dry-cleaners, and more food is bought already prepared for consumption.
[Federici, Silvia (2012-09-01). Revolution at Point Zero: Housework, Reproduction, and Feminist Struggle (Common Notions) (pp. 107). Independent Publishers Group. Kindle Edition.]

Aufheben are incorrect. There is an immense secret being kept - even with generations of feminism from the 1960s until the 2010s and certainly within Marxism - that a huge proportion of the Earth's wealth is generated by unwaged (generally women's) work. Does this work create value? In Fortunati's sense, no. In Marx' sense, definitely.

The Society of the Spectacle

I recently re-read The Society of the Spectacle. I read the Knabb translation.

There is a pdf of the book online and physical copies are available. I needed it for my Kindle, however, so I downloaded the HTML, cleaned up the document and converted it to a mobi format.

UPDATE: Notes from the Sinister Quarter has created a superior version of this book. Go and get it from their website.

Below are some interesting quotes I found during this reading:

Kennedy survived as an orator to the point of delivering his own funeral oration, since Theodore Sorenson continued to write speeches for his successor in the same style that had contributed so much toward the dead man’s public persona.

Wherever abundant consumption is established, one particular spectacular opposition is always in the forefront of illusory roles: the antagonism between youth and adults. But real adults — people who are masters of their own lives — are in fact nowhere to be found.

Like the old religious fetishism, with its convulsionary raptures and miraculous cures, the fetishism of commodities generates its own moments of fervent exaltation. All this is useful for only one purpose: producing habitual submission.

The plain facts of history, however, are that the “Asiatic mode of production” (as Marx himself acknowledged elsewhere) maintained its immobility despite all its class conflicts; that no serf uprising ever overthrew the feudal lords; and that none of the slave revolts in the ancient world ended the rule of the freemen. The linear schema loses sight of the fact that the bourgeoisie is the only revolutionary class that has ever won;

Imprisoned in a flattened universe bounded by the screen of the spectacle that has enthralled him, the spectator knows no one but the fictitious speakers who subject him to a one-way monologue about their commodities and the politics of their commodities. The spectacle as a whole serves as his looking glass. What he sees there are dramatizations of illusory escapes from a universal autism.

The most annoying thing about Windows 7...

is when your application goes off screen and is unreachable. How can that even be possible? I would have thought that Windows would have something to detect and prevent this occurring. I just found a fix, however, so it isn't too difficult to deal with.

VB.NET cheat sheet

I've been doing quite a bit of programming in VB.NET recently. It's almost exactly the same as C# but a few things have caught me out. I've written up a small cheat sheet with the noticeable differences (plenty of websites will give you a huge list of irrelevant differences).

Keywords

C#	VB.NET
this	Me
base	MyBase
abstract	MustOverride/MustInherit
virtual	Overridable
sealed	NotInheritable
class Class : Interface	Implements (statement)
internal	Friend
static	Shared
typeof()	GetType()

If you want a static class in VB.NET, you'll need to use the Module keyword.

(One thing to note here is how much more intelligible some of the VB.NET keywords are.)

Logic

C#	VB.NET
&&	AndAlso
||	OrElse

There is no equivalent for And and Or in C#.

Numeric type suffixes

C#	VB.NET
12.34M (for Money)	12.34D (for Decimal)
12.34D (for Double)	12.34R (for Real)

Lambdas

C#	apples.Single(x => x.Colour = "red")
VB.NET	apples.Single(Function(x) x.Colour = "red")

Initialising lists and objects

C#	var apple = new Fruit { Colour = "green" };
VB.NET	Dim apple = New Fruit With {.Colour = "green"}

C#	var apples = new List { new Fruit { Colour = "red" }, new Fruit { Colour = "green" } };
VB.NET	Dim apples = New List(Of Fruit) From {New Fruit With {.Colour = "red"}, New Fruit With {.Colour = "green"}}

Anonymous types

C#	apples.Select(x => new { Colour = x.Colour });
VB.NET	apples.Select(Function(x) New With {.Colour = x.Colour})

Nulls and Nothing

If you're coming from C#, the Nothing keyword does not do what you'd expect. What would you expect the following code to do?

Dim value = ""
If value = Nothing OrElse value Is Nothing Then
    Throw New Exception()
End If

If you said "not throw an exception" you'd be wrong. Weirdly, the first condition is true but the second condition is false, so it throws. Compare with similar C# code:

var value = "";
if (value == null)
    throw new Exception();

In this case, the exception doesn't get thrown.

Barrett is an idiot

I spent 1.5 hours of my life last week listening to people talk about the "Seven Levels of Consciousness".

Richard Barrett, a man lacking wits but making up for it with entrepreneurial enthusiasm, has improved the old pyramid model of Maslow (the hierarchy of needs) by creating an hourglass model.

Old:

Improved:

One will instantly see the superiority of this new model. For one thing, it's symmetrical. And it has circles. If there is anything more scientific than pyramids, it's circles. And the number seven.

I have to say how impressed I am. Barrett has taken a relatively meaningless concept, the hierarchy of needs - wholly unproven and unprovable - and improved it by extension and inversion via the science of the Vedas. If it's science you're after, an ancient holy text is the best place to look. He's melded religious myth and pseudo-science and made a business out of it, selling it to morons world-wide.

Vedic science specifies seven levels of consciousness. These are waking, sleeping, dreaming, soul consciousness, cosmic consciousness, God consciousness and unity consciousness. It appeared to me that the descriptions of the last four of these levels of consciousness described the underlying features of self-actualisation. (From Maslow to Barrett)

One could wonder how this sort of garbage could become such a integral part of corporate consciousness, especially at the management level. But it's not an anomaly. Business is full of unproven ideas and myth. Modern myth may appear to be especially good targets of ridicule but old myths are no less absurd, merely more accepted because they've been around longer.

Euler problem 19

I started writing this blog entry a year ago. It's about an Euler problem I solved in F#. The code makes more sense to me now than it did a year ago and I haven't touched F# since then. I guess I'm just way smarter now.

---

Solving Euler problem no. 19 is the solution I'm most proud of. It's the first problem that I solved in F# with no assistance by the 'net.

The problem:

How many Sundays fell on the first of the month during the twentieth century (1 Jan 1901 to 31 Dec 2000)?

It isn't an overly complex problem but there a couple of tricky aspects. The approach that I used was to start on the first Sunday of 1901 (6th Jan) and add seven days over and over (i.e., only counting Sundays) until the end of 2000. I could have used the .NET DateTime type to solve this problem very easily, but I decided to see if I could solve the problem using my own date type.

I solved the problem by using a few F# features, namely:

• pattern matching
• tuple
• record
• list

The first problem was February. February is a prickly month. Pattern matching will solve it! The febDays function accepts a year as a parameter and returns the number of days that February has.

    let febDays y = match y with
                            | y when y % 400 = 0 -> 29
                            | y when y % 100 = 0 -> 28
                            | y when y % 4 = 0 -> 29
                            | _ -> 28

After February was ready, I needed to know the number of days in any month, given the month (as a number) and the year. I used pattern matching again. Therefore,

    let daysInMonth (m, y) =
            match m with
            | 2 -> febDays y
            | 4 | 6 | 9 | 11 -> 30
            | _ -> 31

The other interesting function was to be able to add a day to a date. I did:

    let addDay date =
        match date.day with
        | d when d < 1 || d > daysInMonth(date.month, date.year) -> failwith "Not a valid day of the month."
        | d when d = daysInMonth(date.month, date.year) -> match date.month with
                                      | m when m < 1 || m > 12 -> failwith "Not a valid month."
                                      | m when m = 12 -> { day = 1; month = 1; year = date.year + 1}
                                      | _ -> { day = 1; month = date.month + 1; year = date.year}
        | _ -> { day = date.day + 1; month = date.month; year = date.year}

The code above isn't the whole solution, but it's the interesting parts. As you can see, I pretty much pattern matched the whole solution. It's a shame that C# doesn't have pattern matching because it's a really powerful language concept.

Flat Earth vs Climate Change

The following is an exchange between me and someone I'll call Matt. He has some interesting ideas about climate change... After his response, I didn't think I could take the discussion anywhere else. He's clearly a downer and therefore not amenable to reason.

---

Well constructed and very much in line with what I think on the topic, particularly the part about The vested interests and the role of the big ball of burning gas above our heads in global warming.

http://www.theage.com.au/opinion/climate-change-science-is-a-load-of-hot-air-and-warmists-are-wrong-20120801-23fdv.html

Gee I’m glad I get to give an extra 500 per year to the government though.

Matt

---

Matt, I think this debate is distracting you from a much more significant dispute. The real controversy is between those that believe in the Ice Wall and the downers – those believing in the Waterfall. They both have photographic evidence:

Ice Wall:
     







Waterfall:

But whose testimonial can we believe? Personally, I think a British Naval Officer’s account is very convincing:

It would be impossible to conceive a more solid-looking mass of ice; not the smallest appearance of any rent or fissure could we discover throughout its whole extent, and the intensely bright sky beyond it but too plainly indicated the great distance to which it reached southward.

This debate is not simply an Antarctic concern, it has global ramifications that also impacts on the “theory” of global warming. If warmists are correct, the implications of an ice wall is extremely concerning. I think downers are drawn towards their “theory” more because of alarmist fear-mongering than they are by empirical evidence.

It’s even possible that their photo could have been faked!

Patrick

---

Hi Partick,

Agreed, global warming in all of its various extremist incarnations is distracting us all from much more significant debates and issues.

It is far more relevant deciding what to have for dinner tonight and we can all really make a difference here.

In light of my own experimentation and investigation it is just not possible for me to deny that the earth is warming. Just this morning I had to take about 2mil of ice off my car windows. This time last year, it was probably more like 3mil.

My position is more in line with that of the author of this article.

The question that seems to be supressed at every debate on the issue is:

How much is man kind contributing and how much is a natural cosmic or environmental cycle?

The very idea that man kind may not be directly responsible for global warming is considered heresy! In science! According to our politicians - the science of climate, the most chaotic system known to man, is decided, and we're all to blame.

In truth, what makes me most concerned is that people are tired of hearing about it and cannot fathom the thought that perhaps they have been misled to such a gargantuan degree the very world will never be the same.

It all just makes me so HAPppy!

This article was all about the global ramifications of warmist theories and the rampant misuse/misinterpretation of so called "empircal" data so I very much agree with you - individuals will use populist interpretations of statistics to prove whatever wacky theory pops into their heads.

84% of all people know that.

Computer Othello, Part 5: Resources

This post lists the best info I could find on how to write a computer version of Othello.

General

http://en.wikipedia.org/wiki/Reversi#Rules
Rules for Othello/Reversi

http://chessprogramming.wikispaces.com/Othello
Details on bitboards, hashes, deep-first searches and transposition tables. Contains source code (generally in C or C++)

http://www.radagast.se/othello/howto.html
A description of what is required to implement a better than average computer player.

https://skatgame.net/mburo/ps/compoth.pdf
A paper that describes the best Othello computer players from the 80s and 90s (IAGO, BILL and LOGISTELLO)

Strategy (human and computer)

http://radagast.se/othello/Help/strategy.html
http://www.samsoft.org.uk/reversi/strategy.htm

Implementation

http://users.informatik.uni-halle.de/~jopsi/dass4/
A break-down of the tasks involved in creating an Othello game. Has info on implementing the rules and how the minimax search works.

http://www.dcs.gla.ac.uk/~daw/masters-projects/dissertations/Colquhoun.2008.pdf
Computer science student's paper.

http://www.cs.kent.edu/~jmelnyk/othello/
A description of someone's attempts to write-up Othello depth-first searches (alpha-beta, negascout, MTD(f), Multi-ProbCut, etc.)

Computer player

http://samsoft.org.uk/reversi/openings.htm
A list of the standard Othello openings.

http://xenon.stanford.edu/~lswartz/cs221/desdemona_writeup.pdf
Description of various evaluation strategies.

http://othello.dk/book/index.php/Thor_Database
The Thor database was the only archive of Othello games that I could find on the net.

Finally

If you want to see what I did with this information, there is my source code for Othello. It's fairly well written, the UI looks okay and is easy to use and the computer player plays well. I implemented most things you'd do in a world-class computer player. However, you'd have to make it a lot more efficient, if you wanted to take on those players.

It was interesting, frustrating and fun to try to write a decent Othello game. I learnt a huge amount too.

Best films I've seen this year

I don't see many new great films anymore. I watched most of the backlog years ago. Only one or two new noteworthy films are released every year. However, I've managed to watch some really good films this year. They are:

El Norte: About Guatemalans that escape to the US via Mexico. Probably the best film I've ever seen about a typical experience of an asylum seeker.

12 Angry Men: A jury discuss a murder they're to give a verdict on. They're stuck in a hot room debating the various accounts of the witnesses. The film must be every liberals favourite film, but that doesn't stop it from being a really exciting unravelling of preconceived ideas. "Nobody wears eyeglasses to bed."

Night of the Living Dead: Original zombie film. It has a simple but really effective plot. Extremely tense.

The Lady Eve: It's a rom com. There are a few cheesy bits, but it's generally pretty funny and endearing.

Frozen Planet (not a film): A nature documentary series about life in the Arctic and Antarctic. It made me appreciate just how utterly brutal life in the natural environment is for every living creature except (most) humans (and some pets). Nature is cruel, violent, and completely indifferent. Creatures spend their lives hungry, hunted, cold and alone.

It made fully realise that rather than trying to emulate (bourgeois rhetoric on markets), admire (people who like natural/organic products), or lament the loss of nature (millenarian and some Green movements) we should thank our lucky stars that we're somewhat removed from that pitiless existence. We should do everything we can to reject the laws of nature (that doesn't, in turn, threaten our existence).

Hardcore Minecraft

We played multiplayer Minecraft again tonight, after many moons. This time we played in hardcore mode; no re-spawn. It played like the game I've always wanted Minecraft to be. Challenging, tense and satisfying. I definitely want to play more.

A majestic canyon

A field of sheep

Mana from Heaven

My companion's remains. The rock fell out from under him, into lava. He followed.
It was a disturbing, saddening and somewhat guilt-ridden experience.

Troubles with Unity and Mono

Using Unity for my Othello game was not without issues. The biggest issue I experienced related to Unity's use of Mono. Here is why:

A depth-first search generates a lot data. Oridinally, in .NET, that would be fine. You create the data, process it and forget about it - the .NET garbage collector (GC) will release the data from memory when required. That isn't true of Mono 2.6. The garbage collector in Mono 2.6 is kind of rubbish. (See what I did there?) Mono 2.8 has a new garbage collector, but Unity 3.5 uses Mono 2.6. And my Othello game uses Unity 3.5.

For my Othello implementation, every turn by the computer player was a new search. It leaked memory all over the place because the GC wasn't dumping the data. It could easily use a gig of RAM over a game, even with shallow searches (depths of 5 or less).

To resolve this issue, I - though Andrew came up with the idea - used a struct instead of a class for the objects used in the search (EvaluationNode and GameState). I re-used memory by storing search results in arrays where the indexes were reset every turn of the game. This negated the need for garbage collection. In the source code, the classes that manage this process are called the EvaluationNodeBuffer and EvaluationNodeCollection.

These changes turned out to be a really good use and re-use of memory. It is also an excellent example to demonstrate the differences between a struct and a class. It also allowed me to search to much greater depths for the computer player.

A problem with this technique is that it makes it very difficult to write code to re-use part of the search tree between turns. Finding which parts of the tree to prune and to then re-organise the arrays and indexes would be technically tricky and CPU intensive. Therefore, for now, the computer player continues to re-searche all game states between turns. Furthermore, any sort of parallel programming to speed-up the search would be hindered by this approach.

What about Unity 4? That's out soon. Will that support the newer version of Mono? Unfortunately not.

We will be shipping Mono 2.6 with Unity 4.0. This will allow the same subsets of .NET features as in Unity 3, depending on the profile you choose in your player settings. (Unity 4 FAQ)

The Lello-Lea Hypothesis

The Ice-Wall of Antarctica

A report from the Society:

The standard flat-Earth model (SFEM) postulates an ice-wall beyond Antarctica. SFEM supersedes waterfall theories of other flat Earth models and replaces the controversial spherical-earth conjecture's (SEC) notion of Antarctica as an island continent.

The Lello Hypothesis states: "the universe is a series of flat earths stacked within a vast crystalline cylinder folded back on itself to form a never ending torus. I.e., a cosmic, hollow, ice-donut." The Lea Modification, sometimes known as the Lello-Lea Hypothesis, states that: "interspersed with crystalline water (H₂O), exists an abundance of naturally occurring granular material composed of finely divided rock and mineral particulates forming 'deserts' - similar to observable regions on Earth."

It should be noted that these hypotheses are not to be confused with the weaker and frankly implausible set of ideas that A. Hayes has expounded in which a series of flat earths are distributed across the inner surface of a torus of indeterminate extent. Hayes is also known for the infinite cylinder hypothesis (ICH). ICH is perhaps even more unlikely than his inside-out donut.

The Society for the Lello-Lea Toroid Flat Earths Hypothesis is searching for intelligent life on other flat earths. First contact will provide the empirical evidence, and hence proof, of the Lello-Lea Hypothesis. Meanwhile, scientists are formulating a mathematical proof for the minimum quantity of flat earths that must exist.

Chalmers' recent work (unpublished, in correspondence with the author) has yielded a breakthrough on the topic, establishing deep relationships with elliptic curve theory and the calculus of manifolds, while raising the lower bound on the number of flat earths that must exist to -4. We are assured that more revelatory findings are to follow.

It is not expected that the combined forces of the Society and scientists will be able to provide the actual number of flat earths in our toroidal universe. However, we believe that within the next ten years, we will be able to prove that:

1. We do indeed live in a multi-earth, toroidal universe (i.e., SEC is fallacy.)
2. There are at least 3 flat earths.

A. Hayes, and others, have proposed expeditions to the Ice-Wall to conduct experimental drilling. A small party would also test a radical SFEM speculation of going "beyond the Ice-Wall." Funding and applications for strategic and/or technical approaches is being sought.

Computer Othello, Part 4: Trials and Transposition Tables

By the time I implemented transposition tables for my computer player in Othello, I'd finished most of the components of what's required for a decent Othello computer player. I didn't really need to add a transposition table, but I really wanted to understand and implement all the aspects of a computer player for these sorts of games.

A transposition table is used to help speed-up a depth-first search. It does this by keeping track of game states that it has already searched. Along with the game state, it records the value it found for that state in the evaluation function. (Therefore, a suitable data-type for a transposition table in C# is a Dictionary<gamestate,float>. GameState overrides the GetHashCode method to provide an unique as possible hash of the game state. The float holds the value from the evaluation function.) If the search finds the same game state again (via a different path), it uses the value in the transposition table rather than re-calculating it. This saves time.

Intuitively, I wouldn't have thought Othello would have many ways in which different sequences of plays could result in the same game state. Any reading about computer Othello says otherwise, however. In practice, I managed to wipe a couple seconds off a search by using a transposition table.

When I began writing my transposition table I looked for a good way to hash a game state. I found a description of the Zobrist hash:

One useful method for hashing positions in games like Othello and Chess is Zobrist hashing [12]. A Zobrist hash consists of an XOR sum of several bitstrings. For each square on the board, there is one randomly generated bitstring representing a black piece and another representing a white piece. A position's Zobrist hash is formed by XORing together the appropriate bitstrings. The primary benefit of Zobrist hashing is that it can be incrementally updated very quickly by XORing it with the bitstrings for the pieces that have changed. Zobrist hashes also have the advantage of uniform distribution. (Applications of Arti ficial Intelligence and Machine Learning in Othello)

I implemented a Zobrist Hash for Othello, created a dictionary of type Dictionary<ulong,float>, then wondered why I was getting so many bad results. The problem? When you add a key to a dictionary it'll call GetHashCode on the object to retrieve the hash. What happens when you add a hash as a key to a dictionary? The same thing. I was double hashing! Not only that, I was going from a 64-bit Zobrist hash to a 32-bit .NET hash. I was burning time in creating the hash as well as losing information and most likely increasing the number of collisions between different game states. All pretty ugly stuff. The solution, outlined above, was very simple: ditch the Zobrist hash, override the GameState GetHashCode to have:

        public override int GetHashCode()
        {
            return (PlayerPieces | OpponentPieces).GetHashCode();
        }

and change the dictionary to have GameState as the key. The result isn't as fast as if I had implemented my own type of hashtable with my own hash type, but it was a lot easier to do.

In the end, adding a transposition table to my game was relatively easy to do, though I took a long and unnecessary detour to achieve it.

The Walking Dead review

Fashionable zombie pop-culture bores me. I have a negative interest in anything to do with zombies. And yet... there is DayZ and now Telltale's latest adventure game series, The Walking Dead.

The Walking Dead is the only adventure game in the last fifteen years that simply must be played. Why?

• The narrative and plot are compelling
• The script is good
• The characterisation is generally quite good
• The puzzles are a balance between tricky, frustrating and satisfying
• The art style is really good
• The interface is one of the best I've seen in an adventure game

But importantly, it's because

• It appears to have meaningful choices at almost every node of the dialog tree
• What you need to do to survive is a great blend of disgust and grotesque fascination. (I think I've said "Oh my God!" more times playing The Walking Dead than any other game!)

Many dialog trees in the game have a timer on them. You often need to respond in moments. This is the most fantastic innovation in adventures games since the dialog tree was created. I doubt The Walking Dead was the first to use it, but it uses it extremely well. It's so tense. I was often locked in indecision. Occasionally, I simply couldn't decide, resulting in awkward silence. I love how not saying anything is an option! Of course, no matter what you say, it can't matter too much, yet it doesn't feel like that at the time. And that's the important thing.

Often events don't go the way I wanted or how I expected, but I just went with it. It's part of the story and it all seems meaningful. Only once have I re-started a section because of the way things turned out. This is great in two ways; 1) even though things don't necessarily go the way you wanted, it's not necessarily bad enough to force you to rewind and 2) I care enough about some events that I simply can't allow the game to continue after an event that was simply too wrong for me to be happy with. That might sound contradictory, but isn't.

The reason why DayZ and The Walking Dead are such good zombie games is because they're not really zombie games. DayZ is about the other people you play with, whether other survivors or bandits that hunt you. The Walking Dead is all about the other characters in the game. The zombies are there in a way that function almost as a relief. Zombies are simple. All you have to do is kill them (again). It's everyone else you have to be worried about. (But that would make a great twist, if zombies weren't exactly as we assume. If they transitioned from a Nazi SS officer to an Itialian fascist infantryman, then you'd have a whole new depth of narrative to explore.)

Episodal format? Seems like a dodgy way to get people to pay for a game that isn't finished. Episodal gaming is a bit of a relic in the era of kickstarter. It's proto-kickstarter, where some of the risks are put onto the consumer rather than all. In effect, I don't mind at all. In fact, I think I prefer it. I don't have the desire or time to sit down and play a game for hour on end. This way I can play it in chunks over a period of months. At episode two of six, I feel like I've gotten all I thought I would get from this game and I'm not even halfway. If quality drops by the end of the season, I don't think it matters. It's already been a whole lot better than many games I've played recently (looking at you Diablo 3).

This is the first game in years that I want everyone to play. I'm not just talking about my friends that don't play games anymore now that they've "grown up." I mean, my mum. She should play it. Sure, it's horrific, but it's done so horrifically well.

Computer Othello, Part 3: Othello computer opponent

After working on the game archive stats, I was ready to create a computer player. Previously, the computer player had played randomly (even Andrew was able to beat it!) To create a computer opponent, I needed to do two things; create an evaluation method and plug-in a search algorithm to find the best estimated play to make.

Evaluation

One of the trickiest things to do with the computer opponent is to evaluate the current game state. Given any arrangement of pieces, you need to be able to determine which of the possible plays is the best. You need an heuristic to do this. An heuristic is an estimation of what you think is a good play to make. The factors that I used to evaluate the state of the game were:

* Number of pieces
* Number of playable squares (usually called "mobility")
* Number of empty squares next to the player's piece ("the frontier", or "potential mobility")
* Various corner and edge patterns (edges and corners are usually better moves)

Pieces and mobility

The most naive heuristic is to count the number of pieces each side has. Ultimately, this is the only measure, piece count determines the winner. However, until the end game, it has little to do with who is going to win.

A less naive approach is to count how many plays can be made. A state where 7 plays are possible is almost certainly better than one where you have to skip your turn because you can't make a play. I had already calculated where a player could place a piece in their turn as part of the user interface. Therefore, I got the mobility measure for nothing.

Finding the empty squares next to a player's pieces was also relatively simple to implement. It was little more than a variation on the mobility algorithm.

Patterns

Image that elucidates where the X and
C-Squares are on an Othello board.

Implementing the patterns was relatively easy to do but it was one of the last additions to the evaluation function that I made. I had previously dismissed this approach as too simplistic to provide much of an improvement to the evaluation. I couldn't have been more wrong. Once I had coded the patterns, the computer player beat me easily. I've since beaten it once or twice, but it went from a disgustingly below average player to a better than average player.

The patterns that I check for are:

• Corners (a good play for the current player)
• X-Squares (bad play)
• C-Squares (bad)
• Corners and X-Squares (good)
• Corners and C-Squares (good)
• Edges (good)

Furthermore, I check for lots of combinations of these patterns. I.e., not only does a game state get points for a play on a corner, but gets even more points for the multiple corners. I didn't check to see if this approach improved play, but I suspect it would.

Piece stability and parity

There were a couple of major components of a good Othello evaluation function that I didn't include. These were piece stability and parity. Piece stability, i.e., finding the pieces that cannot be flipped, is one of the trickier things to determine. There is a good description of how to do it here. I couldn't think of a really efficient way to implement stability, so I left it out.

Parity, i.e., determining who plays last, was relatively simple to implement in its basic form. By default, white will always play last, and therefore has an advantage. For black to play last, someone has to miss a turn. The basic approach to parity didn't really seem to impact the performance of the computer player, so I left it out of the evaluation. A sophisticated form of parity - one where isolated parts of the board are evaluated for parity (an isolated section is one that is surrounded by pieces and edges) - seemed too tricky to implement, so I never tried.

Depth-First Search

I took my negamax, alpha-beta pruning negamax and negascout search methods from my noughts and crosses source code and adapted it to work for Othello. That was fairly easy to do, although my original code was a bit rubbish.

Initially, I thought I'd use negascout for Othello as it is the best of the three. However, for it to work effectively (i.e., better than an alpha-beta pruning negamax), it needs to do shallow searches of the game tree, or find some other way to have a pretty good attempt at pre-ordering the plays from best to worst. Negascout generally does a mini-search within a normal negamax search. It was a more involved task than I suspected. Once I had implemented the patterns approach to the evaluation function, I realised that my computer player was pretty good. Therefore, I decided not to pursue a negascout algorithm for Othello.

Opening book

With all the work that I did to be able to display the history of games to the human player (percentage of games that made a play, percentage of those where black wins), I was serendipitously writing the code for the computer player. The computer opponent uses this information in a similar way as a human.

End game search

One of the things that I didn't do for the computer player was an put any work into an end-game search. This sort of search is much deeper and tries to get search until the end of the game. Once a computer has this information, it'll know if it has won and exactly which moves to make to ensure victory. Until the end-game search, all other plays are calculated guessing. All I did to the computer player approaching end game was increase the search depth.

Conclusion

I completely underestimated how difficult it would be to create a competent computer player. I now have a much greater respect for people who managed to create computer players that are vastly superior to mine on machines that are vastly inferior to today's technology. It's true that .NET is not really up to the challenge (unmanged code like C and assembler would be much more suitable), but you'd think a modern processor using .NET could get close to Pentium using C. From my experience, it didn't. But in truth, it was the algorithms that weren't good enough. I would have many more things to do to be able to compete with other computer Othello players. E.g., stability detection, parity, better potential mobility, negascout, Multi-ProCut, end-game solving, much deeper searching, training and machine learning etc. There is an immensity of improvements that I didn't even touch on. I'm not unhappy with what I achieved, more very impressed that people have done so much better.

Computer Othello, Part 2: Game archive as trainer

I wanted to create a computer game of Othello that did more than simply defeat the human player. I thought of trying to provide a story to the player to help them learn and improve against others. One idea was to process the Thor archive and create stats to apply to each potential play. In that way, the player can see how many people have played the position in the past and how well they did. However, with 100,000 games in the archive, doing this efficiently was tricky.

Initially, I tried using a trie data structure. It seemed like an appropriate structure. However, performance was poor and programming against it wasn't overly intuitive. I tried changing how the data was stored to try to improve performance. Instead of using a human readable notation (i.e., algebraic notation, such as "E6"), I stored each play as a single Unicode character. With 60 possible plays, I used the alphanumeric set, capital letters and symbols. I did this by simply taking a numeric board position (0 to 63), adding 48 to it (to move it into the alphanumeric area of Unicode) and converting it to a char type. This improved things, but performance using the trie was still terrible.

I ditched the trie and wrote LINQ statements against a list of strings. Performance improved, but it was still poor. I switched the search from LINQ to use the BinarySearch method of a list. For a binary search to work, I had to order the data. I was surprised to find that the LINQ OrderBy method didn't sort my data the way I needed. I suspect that it doesn't distinguish case and can't cope with unusual characters. I switched to Array.Sort. Even that didn't work without the StringComparer.Ordinal option. I also needed StringComparer.Ordinal option for the BinarySearch method. All interesting little hurdles.

After all that (plus some more optimisations discovered by using Performance Analyzer in Visual Studio), performance was massively improved but still not perfect. I decided to process the calculations on a separate thread so that it didn't hang the user interface.

I was done. Stats appeared either instantly or within half a second and either way the user interface was not impacted. Yet, I still wasn't done!

Othello is a highly symmetrical game. The first play, a choice of four possible positions, is symmetrically equivalent. No matter where black plays on the first move, one could rotate the board and the board would look the same as any other play. The people who made the Thor database understood this. That's why they standardised the archive to have E6 as the first play for all games. I needed to display the same info on the C4, D3 and F5 tiles that I displayed on E6. Otherwise, the less knowledgeable player may wonder why 100% of games are played at E6.

To be able to display stats on a non-standardised position, I needed to test for symmetry in the game board. I found methods to flip and rotate bitboards. That solved half the problem. I could check if one game state was the same as another. Next, I rotated the stats from a standardised to a non-standardised board. I generated lookup dictionaries to be able to rotate the indices of the board positions. For example, position E6 would become position C4 on a rotated board.

Display of game stats for black.
The blue highlighted square displays black played there in 0.027% of games and won 10% of those games.

After all that, I had a well working set of stats that I could display to the player to assist them in learning where the best opening moves are. It doesn't really provide any sort of story for the player, but at least it gives an indication of where a good play might be.

Later, I also used this data to help the computer player make its opening moves. But that's for another day.

Julia Kristeva

On Anthony's visit to Melbourne, he mentioned something about Julia Kristeva that I simply had to know more about. It was the idea that Kristeva suggests we move away from the scientific binary logic of false and true (0 and 1) to a poetic logic of 0 and 2. I was astounded to hear that. It seemed completely absurd (it's been a while since I read any post-modernism). I got the document and found the quote. Here it is:

The polyvalent logic presupposing an infinite number of values in the interval false-true 0 ≤ x ≤ 1 is a part of bivalent (0-1), Aristotelian logic. Poetic logic is inscribed on a different surface. It remains indebted to Aristotelian logic not in being a part thereof, but insofar as it contains and transgresses that logic. Since poetic unity is constructed in relation to an other as double, the problem of truth (of the 1) does not concern it. The poetic paragram bypasses the one, and its logical space is 0-2, the 1 existing only virtually. (pg 44, Towards a Semiology of Paragrams)

Admittedly, I didn't read all of the document. What I read, I didn't understand. I'm taking Kristeva out of context because I don't understand the context. Nevertheless, I got questions:

1. How is polyvalent logic "part of" bivalent logic? The inverse (bivalent part of polyvalent) seems more reasonable to me. Yet, polyvalent and bivalent seem to have quite different logical rules.
2. Unlike polyvalent logic, Kristiva claims that poetic logic (what's that?!) has a logical space of 0-2. It's therefore still bivalent, but because it doesn't contain 1 (truth) it need not concern itself with truth. Hm... but isn't the "1" merely a symbol to represent "truth"? Couldn't "2" also represent "truth"? Or "0" for that matter? In fact, you could be really crazy and suggest "√-1" to be the symbol for "truth". It wouldn't matter, would it?

I will leave you with another quote from earlier in the document. I find it entirely incomprehensible, though it does appear to contradict the above by saying that paragrammatic numerology (which I think is the same as poetic numerology) is "two" and "all" rather than "0" and "2". I like the "The zero is two which are one" statement, though that is very similar to another phrasing I know (i.e., the doctrine of the trinity - god, son and holy ghost are three that are one.) I'm not sure that the bible is a good source for learning logic, however...

The zero as non-sense does not exist in the paragrammatic network. The zero is two which are one: in other words, the one as indivisible and the zero as nothingness are excluded from the paragram, whose minimal unity is both an (empty) all and an (oppositional) two. We shall examine more closely this paragrammatic numerology, where there is no 'one' or 'zero' but only 'two' and 'all'. Unity is empty, does not count, the one is zero but it signifies: it controls the space of the paragram, it is there to fix the centre, but the paragram does not give it a value, a stable meaning. This 'unity' is not the synthesis of A and B; but it has the value of one because it is all, and at the same time it cannot be distinguished from two, because within this unity come together all the contrasting semes, both opposed to each other and united. At once unity and couple, the oppositional dyad, to apply a spatial expression, is realized in the three dimensions of volume. (pg 37, Towards a Semiology of Paragrams)

More interview questions

We had a coding competition at work last week. We had an hour or so to code-up a solution to:

I want to simulate a vote by providing a question, valid responses, and an integer representing the voting population size.

When the vote occurs, there should be 1 random vote for each member of the voting population.

 The entries were judged on the following criteria.

• Testability
• Maintainability
• Accuracy of solution
• Structure of solution files

I've uploaded my contribution to GitHub. I didn't win. I made a few weird assumptions (that I've corrected in the code online) and for some reason went off on a tangent playing around with JSON serialisation. However, the wining solution wasn't particularly different to mine.

The purpose behind the competition was to use this task (or something similar) when interviewing candidates. I guess the people interviewing wanted a base-line to work off, that's why they had the staff do the task.

Anyways, it was quite fun. I've never done anything like that in a workplace before.

Interview questions

I've been asked the same questions during interviews in the past. The two that come to mind are:

1. What's the difference between a clustered and non-clustered index in SQL Server?
2. What's the virtual keyword in C# mean?

I finally got around to looking up the answer to the second question. I generally don't override methods, so I've never had much use for it.

The short of it is that virtual is required if you want to override a method of the base class. If you don't supply virtual, the best you can do is use the new keyword to create a completely new method that just happens to have the same name of a method in the base class. To take it a bit further, sealed will prevent any further overriding down the inheritance chain. Therefore;

1. a base class may have a method with virtual assigned to it;
2. deriving from 1, you could override the method;
3. deriving 2, you could sealed override the method;
4. deriving 3, you can't override any more (but you could new the method);

Is this an important question? Not in-itself. I was offered the jobs without knowing the answer.

I have written an example gist to further clarify.

GitHub for Windows

I'm tired of using Skydrive as a "source code control" solution. That's no solution, just a file back-up. What about versioning and sharing code?! I've used lots of revision control systems before. Here's the list:

• Visual SourceSafe
• SourceGear Vault
• Subversion
• TFS
• Mercurial

They've all been adequate but frustrating. The Internet is abuzz with talk of Git and GitHub. I tried setting up a GitHub account last month. After the first few steps, I looked at the long list of remaining tasks and decided it was "too much work."

To my rescue comes GitHub for Windows. Not only did it create my repository in seconds, it filtered out all the useless gunk that I don't want going into the code repository, namely binaries, ReSharper and Visual Studio profiler files. It still had some NCrunch files to ignore, but that's a minor issue. Could this be my revision control Holy Grail?

My Othello repository can be found at: https://github.com/ledpup/Othello.