Covered in Scorpions

May. 31st, 2014

10:29 am - 4.8 on the Bechdel scale

Recently there was this thing purporting to measure relative sexism in Dr Who, between writers.

It struck me that the Bechdel test actually does a pretty lame job of measuring anything meaningful, because it simply assumes that men are strongly represented. As a simple flip-switch (as it was originally written) it's fair enough, but you can't validly draw conclusions from it, and it's especially poor for the purpose given in the article, measuring relative sexism.

To explain what I mean, I'll give my alternative; instead of (as the article does) measuring screen time and talk time for specific characters, we should measure all the characters' conversations. The classic Bechdel says we must have two female characters have a conversation with each other that's not about a man. A simplified expansion on this then, I say we should also check whether we have two male characters that have a conversation with each other that isn't about a woman. I suspect that many Doctor Who episodes would pass neither of these tests; if so then that's not sexist, it's just stories that aren't conversation-driven (or mostly involve conversations between two or more genders).

I'd also suggest some sort of exceptions for title characters. You can hardly call the writers of "Mr Magoo" sexist for most of the show revolving around a guy, it's clearly the focus of the show. That would be a totally separate measure of societal sexism, how many shows are based around males versus based around females. (How do we score Mrs Doubtfire on this metric? Maybe one point for LGBT?) If we were to roll that rule in on the Doctor Who measuring then my guess is the results would make it look like Russell's Who episodes were disjointed and rambling (but with female characters talking to each other!), whereas Moffat's episodes have strong focus and most of the time actually involve the Doctor. What a sexist bastard, strongly featuring the title character in a show he's writing. String him up.

Feb. 17th, 2014

06:42 pm - not selling magazine subscriptions

Yesterday a kid came to our door "not selling magazine subscriptions or anything, just a few minutes of your time", so I gave him a few minutes during which he told his sob story (he's a foster kid, the government is paying his way through college and all his necessities, but if he earns any money during that time the government takes it), leading into, basically, he's selling newspaper subscriptions. As soon as I spotted that's where it was going, I interrupted and said I don't want to waste any of his time, and I can tell at this point I'm not going to be interested in what you're offering. And then he was a douche about it, which frustrates me because if I'd just been mean in the first place, said "not interested" and shut the door in his face, I'm sure he wouldn't have had any mean-spiritedness towards me at all. So I get thought ill of for being nicer than average, because I was willing to help out if he was doing a survey or something similarly harmless like he told me at the start.

But that's not really what aggravates me. What really aggravates me is that the kid is being exploited by some shitty company almost certainly lying to him about how they're going to do something for him if he sells their impossible to sell garbage. Which almost certainly amounts to, given how hard it will be to sell newspaper subscriptions, he's being paid way less than minimum wage, which in any context other than horrible high-pressure sales and lying that nobody wants, is illegal. Why the hell isn't this crap illegal? I don't have much ill-will towards the kid (the being a douche about it is annoying, but I'm sure he's had a horrible frustrating day trying to earn an imaginary buck so I'll cut him some slack), but I would absolutely vote for painful deaths for anyone pushing that bullshit onto kids. Whether his sob story is true or just part of the high pressure sales tactics he's been trained in (and I suspect the latter because he was clearly reciting a rehearsed script), either way it's utterly reprehensible.

Jan. 4th, 2014

12:41 am

I've just learned a valuable lesson about running a business; always be two businesses. Then when people phone to try to sort something out because you've screwed them, put them on hold for two hours with intolerable music and then say "oh we can't help you with that, you need (other business that is really the same business), here's their number." Do that with both numbers, and you never even have to actually provide a service at all! For example, be KLM partnered with Delta, or be Delta and deal with bookings done through Expedia, or a be a hotel and deal with bookings done through hotels.com, or even make Walmart and walmart.com have separate phone numbers and blame each other - it's so easy to be incredibly unhelpful!

Make sure to have both customer service and your labyrinthine phone menus (which must never make any difference to who you end up speaking to nor provide them with any information you made the customer enter - the customer service staff must make this clear by asking for all the information again) repeatedly suggest that the customer try to resolve their problem at the website, and when the customer has escaped from the phone menu maze you should also interrupt the hold music every 30 seconds to loudly suggest that the customer should use the website to resolve their problem. For bonus points use a variety of different recorded voices so that the customer can't be sure that someone hasn't picked up this time without devoting all their attention to your timesink - if they can just read a book how are they going to be persuaded not to try again? Do not use a recorded voice to suggest how long it will be before someone will answer - mystery is even more frustrating than lies.

Similarly, if a customer tries to use the website it should suggest that they phone, but not give a phone number unless they first navigate a byzantine 'FAQ' of questions nobody would ever ask, confirming repeatedly that the suggested answers are unhelpful. The phone number at the end should be incorrect. If a customer tries to write a complaint or ask a question with a web form, the site should respond with a vague "failed to send message" error, or, even better, a blank screen so the user doesn't even know that the message won't be delivered. Alternatively, declare that the message has been delivered and will be responded to "shortly". Do not give an approximate time as that will encourage people to try again after that time, and do not actually deliver any message.

Your website should, of course, work just fine up until the point at which you've taken money. The customer should feel secure in making a transaction until it's too late, only then do you refuse to give them a refund because "oh someone else is in charge of that, here's their number."

Jun. 23rd, 2013

04:32 am

Someone asked the other day about top ten favorite movies, and I've been slowly coming up with answers since. In no particular order, mine settled on something like:

Which isn't really a true top ten, it's a bit biased towards variety (otherwise there'd be more Stephen Chow) and may also be a bit biased towards movies I would recommend, thus away from movies everybody's seen anyway (ie. there's no Pixar movies in there, though obviously Fight Club made it on despite that bias).

I'd be interested to see other people's (similarly biased?) top ten lists, to treat as maybe-recommendations.

May. 3rd, 2013

08:05 pm

Real-world economics in action! Goozex, a video game trading service that's now pretty much defunct, ran like this: a game has a value in 'points'. You can offer a game for its point value, or you can request a game for its point value. When there is both an offer and a request, the two are matched up, the points change hands, and the requester is also charged a "token", which costs a dollar. The seller pays for the shipping to the buyer, and ships the game. Also you can buy points.

For a while this model worked pretty well, games traded regularly, people were happy with it. Gradually, the request queues got longer, and there were no dangling offers to be found. And in talking about this, and what could cause it to have collapsed like that, I think I found the answer. And it's kind of interesting.

Imagine on a small scale, let's say there are 6 people. One of them buys 5000 points and 5 tokens. Each game costs 1000 points. The other 5 people all have a game they want to trade, so they sell their games to the first person. Now they each have 1000 points and the first person has no points and 5 games. Now in the system is 5000 points worth of demand, and zero supply. The guy who bought the games finishes them, decides to keep one, and sell the others. Now there's 5000 points worth of demand, and 4000 points of supply. The other users buy tokens so they can spend their points, and one of them buys a thousand points too. That one buys two of the games, two of the others buy a game, and two are left with their 1000 points and are in a queue. Now the original user has 4000 points, two users have 1000 points each, and there's no supply again.

This sort of trading can go on indefinitely, the point is, the demand never ever goes down - there will always be 6000 points worth of demand, or more if someone buys some points again. Supply can go up if people buy games for money and sell them for points, down when people make their trades, or stay the same when people make trades and then want to resell the game back in again. But anyone who has points won't want to buy games for money and sell them for points - they want to use their points to get their game, obviously, because cash is good for other things but points aren't. Demand can only ever go up, short of people dying with a points balance and no request queue. So it's inevitable, with this design, that the Goozex economy would fail. The prices were fixed, at first, but now they freed them up which has resulted in massive inflation as the prices go up to try to compensate for more demand than supply, but since there is essentially infinite demand, inflation can't fix it - all it does is make people mistrustful of the system so they try to cash out, making the problem even worse.

The funny thing is they could have solved this problem at the beginning, if they'd thought it through and realized this would happen - rather than charging a dollar for a trade token, they could have just charged a percentage of the points involved in a trade. That way the demand and supply would both 'soften' when a trade happens, preventing their current situation where there is literally no supply and nobody wants to provide any supply - it doesn't matter how many points you can get for your game if there's nothing you can spend points on.

On the other hand, maybe with a points-only system a different problem would arise - people would try to trade in games for more points than they could buy for the same amount of cash, and would refuse to pay cash for points because they would perceive that they could get a better deal. However, this problem would resolve itself - an oversupply of cheap games would reduce their point price (before trades are made since not enough people are buying those games for points) until that game's point value corresponds to the cash value of buying that many points. So in the end there would be no advantage to trading for points rather than buying them, unless you were trading games people actually want, in which case the system is working as intended!

There might be a problem persuading people that buying points is ever a good idea though, when they could just buy the game they want in the first place for the same amount of money. Tokens are a clever way of sidestepping that psychological issue. Alternatively, if the point value (including the transaction 'tax') is always slightly lower than the cash value of the game, then it would make sense to buy points if you don't have any, while still making sense to trade games in.

Apr. 17th, 2013

05:24 pm

I just noticed a really obviously stupid thing about operating system design, mostly Windows but partly true in others as well. There's the concept of the "administrator account", that enables installing software, to prevent things from secretly installing malicious software. But here's the problem - every time we intentionally install something, we give someone's arbitrary program the permission to run as an administrator.

So basically every piece of software we ever use, at the very first point in its life cycle has administrator privilege. At that point, what good is that barrier even doing? I suppose it's useful for preventing buffer overflows and things from giving system-invading access, but those things are a tiny minority of infections - the usual vector is people installing something that has a malicious thing piggybacked on it. That malicious thing now has administrator privileges if it wants them, because it can grant itself them during the install!

It would make much more sense to have a single operating-system-owned "installer" program, and only install packages, globs of files with coded installation instructions. There would still be an annoying "are you sure you want to install this?" popup, and there would still be the possibility of installing malicious software that you might run at the user level, but there would only be an "are you sure you want to give an arbitrary thing administrator privileges?" warning if the installation package was specifically requesting that. The installer program could also have a separate warning for "are you sure you want to install a thing that will run at startup / immediately?" which would vastly reduce the risk of malicious software infections, since there isn't a lot malicious software can do if you have to actively elect to run it every time.

As an added bonus, this would warn you about Adobe and Sun's auto-updaters being jerks before you installed them, too.

Mar. 15th, 2013

02:56 pm

Followup to my Facebook fiasco - I have now got back in by giving it Jessica's phone number. Somehow that is valid confirmation that I am me, while actually valid photo ID went completely ignored for three days. Well played, Facebook. (I then immediately deleted the phone number from my account because fuck you Facebook I don't want a phone number on my account, let alone someone else's phone number!)

Mar. 13th, 2013

04:43 pm

That's pretty messed up - a few days ago Facebook decided to accuse me of not being a real person, and challenged me to identify 5 friends from randomly selected pictures. That would be easier if half the pictures weren't indistinguishable baby pictures, and if Facebook hadn't aggressively encouraged me to 'friend' everyone I've ever met in the slightest capacity many of whom I have no idea what they look like today since I last saw them 20 years ago, but I managed to barely defeat the challenge, and regained access to the account.

24 hours later it decides to accuse me of not being real again, and this time it wants a phone number to confirm that I'm a person. So I give it a phone number. "We're going to send a text now, okay?" Well, no, that phone number can't receive a text, and I don't have one that does. "In that case just scan and send us some government-issued photo ID!"

What the hell? This isn't a high-security dealy like a bank account, it's Facebook, and there wasn't any valid reason for the accusation in the first place - the only reasons I can conceive of for this happening are either that I have a funny name that has matched some new no-fake-names algorithm, or someone has decided to report my account as a fake and Facebook just arbitrarily takes someone's word about such things a second time even after the accusee has jumped through hoops to show the accusation to be false less than a day earlier.

But it's worse than that, because Facebook has become such a ubiquitous thing that many sites have a "log in using Facebook" button - so Facebook deciding to randomly cut you off from your account isn't just cutting you off from their service, they're cutting you off from an unknown number of other services too.

And it's worse than that too, because the remedy "send us photo ID", which I'm willing and able to do because they do say "obscure any parts that aren't relevant", and I have a scanner and am okay with Photoshop (what the hell would my mother in law do with this situation?) ... this remedy isn't actually processed in a timely manner, so even if you're willing and able to jump through hoops you're still cut off from whatever accounts you use Facebook to log in to for an arbitrary amount of time.

Yet another reason why I wish "log in using Facebook" buttons were replaced with "log in using singlepassword.com", a hypothetical nonexistent service for which you would create an account as anonymous as you like and use it to log in to any other accounts.

Mar. 9th, 2013

12:57 am

A product that should exist; robot simulator. With several environments and components as plugins so they can be easily added, with everything based on things that actually exist - so you could build a virtual robot, for example, made of an arduino board or a raspberry pi, say, with moisture sensors or GPS or cameras or LEDs or lasers or motors or arm-controllers or touch-screens or whatever, attached to whatever pins of the boards, plus batteries or solar panels or etc. then you can program it in a virtual arduino/pi programming environment, and see how it operates in a given context.

Uses:

  1. Prototype a concept without having to buy parts or solder anything.
  2. It would be a fun game to make fake robots fight each other or solve problems. It could be better TV than "Robot Wars" because there wouldn't be any safety limitations! (You could apply price or weight limits for different robot classes.)
  3. Get your prototype actually built since the virtual one is essentially assembled from real parts.
There was a game kind of like this years ago, but the programming part was a very limited pseudo-language of drag-and-drop instructions, and the available components were also very limited (there was no GPS or triangulation facility, nor any sort of 'out' signals, only passive 'receive' signals, geared entirely to fighting robots in a very limited arena, rather than solving problems).

I realize it seems like an insanely complicated thing, but there's already basically all the physics simulation that's the hard part, making realistically inaccurate sensors is mostly trivial by comparison (though doing things like lasers flickering at an invisibly high frequency coupled with light sensors or large volumes of flowing water could be tough - it wouldn't be that hard to simulate adequately but it would be hard to simulate in real-time. But for non-human-controlled robots that's fine, you could simulate it slow overnight and play it back real-time the next morning to see what happens.)

It seems like if such a product existed it would likely lead to vastly faster development of all sorts of useful automation. A lot of people like to solve problems or make cool things, but we can't all afford the hardware to experiment.

Dec. 8th, 2012

12:27 am

I am surprised. I was predicting that some time in the last few weeks there would have been a lot of noisy news about poor sales figures over the Black Friday and Cyber Monday time, but I haven't seen a one. I was planning to be grumbling "maybe your sales figures wouldn't be rubbish if you were having actual worthwhile deals!"

Maybe they successfully sold a lot of things despite all the deals being pretty pathetic this year, in which case I suppose we can expect every future Black Friday to consist of ads saying "it's Black Friday! Buy something at its usual price or even more!" And people queuing up overnight to do so just out of a sense of tradition.

Navigate: (Previous 10 Entries)