Interesting Links for 03-12-2011

[andrewducker]

• European Commission proposes splitting accountants' auditing and consultancy work to prevent conflicts of interest
  (tags: accounting )
• Steven Moffat on the rumoured forthcoming Dr Who movie
  (tags: drwho StevenMoffat movies tv bbc )
• Scotland has warmest November on record
  (tags: weather scotland climate )
• Snow and ice forecast for parts of Scotland

Comments

[artkouros]

When I was in 7th grade, my math teacher, sensing my boredom, gave me my first Heinlein paperback. I spent the rest of that year reading them in her class. By the time I finished 8th grade I had read them all. They last one I read in college was The Number of the Beast. I enjoyed the Troopers movie - it was good pulp.

[philmophlegm]

I'm an SF fan who has never read any Heinlein (which is shameful). I have several books sitting in my library waiting to be read, including most of what I suppose are his more famous works.

Apart from the obvious, famous ones, which of his works would you especially recommend?

[andrewducker]

I'm not sure it's shameful. I grew up on Heinlein, because my parents had a collection of SF novels, but there are so many books in the world that it's easy to have missed any one author.

Ones that I'd happily recommend (having not re-read any in a good 15 years):
Double Star, Have Space Suit Will Travel, Starship Troopers, and The Moon is a Harsh Mistress.

[artkouros]

If you just want to read one, read Stranger in a Strange Land. If you want to read several, he has chronological series, see http://en.wikipedia.org/wiki/Future_History , of these The Man Who Sold the Moon and Coventry stand out in my memory.

[octopoid_horror]

I'm kinda curious about the Starship Troopers remake since the Verhoeven film is one of the best sci-fi movies ever made.

[octopoid_horror]

I mean, seriously

[garunya]

It's afraid!

[strawberryfrog]

I'm not entirely on the same page as you.

I loved the book of Starship Troopers when I was a kid reading every bit of SF that I could get. But looked at soberly it glorifies the military with no real logical argument behind that; and has a very strong theme of destructophilic gratification. At least RAH had been in the miliary, so he could write about it with some echo of reality.

The movie is slick, camp and superficial, and often stupid. No military accuracy at all. Maybe that's the point; I can't tell.

There's a good movie in that book somewhere; Verhoeven's movie definitely isn't it. I've been watching Saving Private Ryan this weekend; something between that and Aliens, with a touch of BSG and Gundam would be better. It would still be full of jingoistic explosions but hey, everyone's got faults.

[philmophlegm]

I would love to comment and express an opinion on the audit proposals, but I'm afraid I'm really not allowed to.

[andrewducker]

Because you work in the industry?

[philmophlegm]

A little bit more than that. I work for one of the Big 4 and my job is to coordinate the audit quality review programme of that firm's European practice. I know the private views of some of the key figures in that audit function, and while they're very interesting, they've also been given to me in strict confidence.

[andrewducker]

Aaah, gotcha. Yes, you _really_ can't talk about that!

[undeadbydawn]

so the radical solution to people being in unpayable debt is to give them the means to pay the debt.

Oddly, I'd thought that was the most obvious thing that could possibly be done. Giving money to the banks always seemed to be a monumentally silly waste of cash. Which, we now know, it was.

[I took a £6k loan a few years back, when I was earning enough to easily pay it off in a few years. I then lost my job. It is severely unlikely that debt will ever be paid. it may as well just not be a debt, because I effectively ignore it.]

[andrewducker]

Well, theoretically a lot of the money given to banks was loaned rather than given, but when we're we're writing off the debts of banks (see, for instance, Greece), the idea sounds reasonable.

[undeadbydawn]

loaned with no provision to ever pay it back, as I recall.

Greece is one big anomaly, and I'm nowhere near well enough informed to argue that point.
what is clear is that the speculators are now betting on the success of the bail-out strategy, and in doing so actively preventing the strategy from working. So in effect we have the same people who caused the crash causing another one by continuing to dick around.

this is possible purely because they weren't bankrupted and thrown in fucking jail the first time.

[yes, my arguments suck. I'm tired and fed up :)]

[fub]

Interesting that they would risk using neural networks for risk detection -- I'd think they would want something that is more explainable.
When I was finishing my cognitive science degree, I specialised in knowledge engineering. My supervisor was fond of saying that knowledge engineering was actually quite easy, you just take one of three paths:
- If you know exactly how humans solve the problem, you encode that in a program, and you're done;
- If you have a vague idea of how humans solve the problem, you encode that in a program, and call it a heuristic. If it's right most of the time, everybody's happy and you're done;
- If you haven't got the foggiest idea of how to tackle this problem? Then you use a neural network -- it'll always give some kind of output!

[andrewducker]

I assume that in this case they don't know how to do it nearly as well as the neural nets do!

[fub]

That's a given -- otherwise you wouldn't use a neural net. The biggest drawback with neural networks is that it is very, very difficult to know why your 'net makes some decision. There are other machine learning techniques (my former boss used genetic algorithms) that are much more 'explainable'. That means that you can check your systems -- because you never know what a neural network is picking up on, and, depending on your data set, it might be the wrong things!
See also here.