As soon I say this at work, someone will pipe up that it would be better to count semi-colons or get into a discussion about how bracketing style can affect the count. My experience is that bracketing style doesn’t make much difference, maybe 10%. Semi-colon count can be quite different, but is it anymore meaningful?

There are many more sophisticated measures of complexity. For example McCabe’s cyclometric complexity, which counts the number of paths though the code. And Halstead’s complexity measures, which measures the number of operators and operands in the code.
However I recently read about an empirical study which shows that all of these complexity measures are highly correlated with lines of code. The conclusion is “Whether based on program structure or textual properties, the metrics do not provide more information than simply ‘weighing’ the code by counting the number of lines.”

A couple of rules of thumb. Norris’ Number is 1,500 lines: “The average amount of code an untrained programmer can write before he or she hits a wall.” From my experience at about 10,000 lines of code you have to start introducing some big architecture into the system to partition the code. Windows XP is estimated to have 45,000,000 lines of code.

It is worth stressing that lines of code, or any of these other metrics, measuring how big the program is. This is an indicator of how difficult it will be to learn, maintain or change the software. They do not measure well how much the software “does”, so they are not a good indicator of programmer productivity or program functionality.

It does raise a bit of a problem though; I’m worried about not getting the best deal. I have just learnt a strategy which copes with this bizarre behaviour. I read about it in The Perfect Swarm by Len Fisher, though he says he got it from Simple Heuristics That Make Us Smart.

When you’re shopping for a bargain, and can’t go back and choose a previous item:

• Set a maximum number of things to look at.
• Look at the first 7% of items, making a note of the price but not buying.
• Continue browsing, and choose the next thing which has the lowest price so far.
• This heuristic gives a 92% chance of buying an item in the cheapest 25%.

If you have more stamina, look at the first 14% then choose the next one with the lowest price so far. This gives an 84% chance of buying a item in the cheapest 10%.

I have another rule to protect myself after purchase. The ignorance is bliss rule – after buying something do not look at any more prices. Definitely don’t go in any shops selling the same sort of thing, and preferably go straight home. I never take anything back, so it’s better not know that a better bargain may exist.

In “Keynesian economics” developed in the middle of the 20th century, unemployment is viewed as involuntary.

…employment was determined by aggregate demand for goods and services. This meant that if aggregate demand were too low, production and income would be low, and unemployment would be high. [Backhouse p118]

A major criticism of Keynesianism, is that it is a purely macroeconomic theory; it considers the economy only at a large scale. There was a desire to make more rigorous economic theories based on microeconomic principals. That is, to explain the behaviour of the economy as a whole based on the aggregate action of individuals. This work began in the late 60′s and has continued to the present day.

One of the first of these theories was “New Classical Economics”. In it unemployment is viewed as voluntary

…unemployment was voluntary – the result of workers choices. In their model, unemployed workers were ‘persons who regard the wage rates at which they could currently be employed as as temporarily low, and who therefore choose to wait or search for improved conditions rather than to invest in moving or changing their occupation.’ [Backhouse p130]

What was surprising to me about both these views of unemployment is not the rightness or wrongness of the statements themselves, but rather that they are both assumptions of their respective models. This means that though both theories can make predictions about unemployment, neither can actually tell you why someone is unemployed – they have both assumed that from the start.

• You’re Ignorant
  • The reason you disagree with me, is because you don’t know the necessary facts. You’re ignorant. If only you learnt some stuff everything would be ok. We carry on talking until I realise you do know the facts, at which point I think
• You’re Stupid
  • You know the necessary facts, so the reason we disagree is that you can’t process the information. You’re stupid. If only you weren’t an idiot everything would be OK. We carry on talking until I realise your not an stupid, at which point I think
• You’re Evil
  • You know the necessary facts and your not stupid, and yet you still reach an utterly perverse conclusion. You must be evil. Active malevolence can be the only explanation.

Well maybe it’s not a perfect description of my though process. Normally I’m thinking I got flustered and didn’t explain myself very well. But all the same. A great quote from the video:

What does it feel like to be wrong? It feels like being right.

Checklists are loved by pencil pushing bureaucrats who must detail every minute step of some mundane task. They stifle creativity and are an affront to professional judgement. Real professionals rely on their skill and daring to get the job done.

In his book The Checklist Manifesto, Atul Gawande works hard to dispel this attitude and make the case increased use of checklists in a professional setting. I think he has a pretty good argument and even if you are sceptical, the book’s a good read.

The main thing to realise is the checklist is not a procedure; it does not define exactly how to do the job or detail every step required. Rather it is a summary, designed to catch common mistakes, which even experienced practitioners make. And a technique to foster teamwork and cooperation – to make sure people are talking to each other.

You can see both these aspects in the Safe Surgery Checklist Gawande developed. Many of the items are designed to trap routine mistakes, for example mislabelled specimen bottles. But some are about teamwork – just prior to skin incision the team introduce themselves by name and reviews anticipated critical events in the surgery.

Developing a good checklist is difficult. There is an overriding need to keep the checklist short, or it will not be used. But this means leaving out information. Deciding what to put in and what to leave out is tricky, and can only really be judged by testing and refining the checklist.

In summary a good checklist should:

• Define clear pause points at which checklist should be used.
• Pause points should last no longer than 60-90 seconds.
• Be either Read-Do or Do-Confirm.
• Have between 5 and 9 items.
• Have simple and exact wording, using the common terminology of the profession.
• Fit on one page, free from clutter and unnecessary colour.
• Be Tested and refined.

I installed Akismet to do spam filtering, but this did not deal with the existing spam mountain. I made various attempts to scale the spam peak, and recover the odd real comment which may have been lost there. But this resulted in me maxing out the CPU on the webserver, and my host suspending the website.

After a courteous email exchange JustHost graciously reinstated my account, and I was left feeling somewhat nervous. After googling for a while the only apparent viable option was to delete the comments directly in the database.

After backing up the database and looking at the database schema, I chose the following sql delete commands:

delete from wp_commentmeta limit 1000;

delete from wp_comments limit 1000;

The “limit 1000″, means only 1000 records are deleted at a time, so the commands had to be run repeatedly. First I deleted from wp_commentmeta, then wp_comments. I left 15 seconds between each delete, so as not to hit the CPU too heavily and incur the wrath of my host.

All the comments are gone. Barry and the guy in China — sorry.

I have been watching televised roulette on Channel 5, and have come to the conclusion that it is in violation of Section 5.7 of the Ofcom Broadcasting Code.

The Hot Numbers section of the screen shows the last 9 results of the wheel spin. These numbers are clearly facts about past performance, but they are not balanced by a statement about future performance. On their own they suggest to the viewer that a pattern can be detected which could allow the next result to be predicted. Of course the outcome of the next spin is entirely random, and not influenced by past results in any way.

The Hot Numbers should carry a warning to make future performance clear; something like “the outcome of the next spin is not affected by previous results”

• Post-It notes – I like the traditional yellow. For me they have a greater versatility than electronic counterparts – after all I just want to jot things down, not write a book. The other problem with electronic note taking, is message-medium confusion. By the time I’ve started up the note taking machine, and opened the little app, I’ve forgotten what I want to write. It’s even worse if it’s a bit of math or a little picture. Even more ceremony. I can also stick the Post-It’s into the analogue book I’m reading.
• Pencil – mechanical obviously. My current favourite is a Faber Castel GRIP PLUS 0.7. I find the 0.7 mm diameter graphite more satisfying to write with than the more common 0.5mm pencil. I also prefer a 2B lead, but they are difficult to buy so I normally settle for HB. The Faber has also got an excellent eraser.
• Dice – for when I need to randomize my output. Humans find it almost impossible to think of random numbers, but often the optimal strategy is to behave randomly, hence the need for a portable random number generator.

1) There is no evidence of a connection between MMR and Autism.
2) There is evidence of no connection between MMR and Autism.

Do they mean the same thing or something different?

In fact what exactly do these statement mean? This may seem like a trivial question, I have certainly spent most of my adult life thinking the answer was obvious. But it has recently dawned on me that this is not so. There are in fact two subtlety different definitions of “evidence”, and which one you adopt has a profound effect on the meaning of the statements.

In Cambridge evidence is defined as “one or more reasons for believing that something is or is not true”. It is data on which to establish truth or falsehood. This is certainly how I think of scientific evidence – it’s the collection of observations, experiments, and data which you use to decide if a proposition is true or false. With this definition statement (1) means “no data exists, which could be used to determine if there is a connection between MMR and Autism”. Statement (2) means “data exists which shows there is not a connection between MMR and Autism.”

In Oxford they have a different definition of evidence “the available body of facts or information indicating whether a belief or proposition is true or valid: the study finds little evidence of overt discrimination.” In Oxford evidence is only evidence when it purports to the truth of a statement – data about the falsehood of a statement is not evidence at all. With this definition both statements (1) and (2) mean “data has been collected, but it shows there is no link between MMR and Autism.” This is certainly a common usage of the word evidence, see for example [1], [2], [3], [4].

When I think about it, I find the Oxford definition of evidence utterly bizarre. Why should it matter if the data confirms or denies a proposition? Surely it’s all just evidence. It is in fact so stupid I cannot conceive that anyone really believes this definition, it’s just a slightly sloppy dictionary definition.

The problem is people, myself included, keep saying “there is no evidence that….” It just sounds right. It is unfortunate that it is so wrong – in almost all cases there is in fact heaps of evidence.