Category Archives: Accuracy

Data accuracy issues

Real Estate Websites Suck: Part 4

I’ve decided that I’m only going to look for properties with 4 (or more) bedrooms. I enter this as a search criteria, and the website says quite clearly “Results for properties for rent with 4+ bedrooms in {suburb}”.

So why do I get presented with 3 bedroom properties?

Facepalm. Five years, and these web sites still suck balls. Not only do searches not work, it appears that the site pegs my CPU at 100% when the rendered page is just sitting there. Some of their lovely JavaScript goodness I suppose.

If you ask nicely I might dig up and dust off my rant from five years ago…

Weird bug

Let’s say, for example, that a system supplies you the time of some event in UTC, you convert it to local and shove the date/time up on the display. Say, for argument’s sake, you also include the Day Of Week, ending up with a format of DD/MM, DOW HH:MM. Everything looks fine, until someone notices that the Day Of Week is wrong. The 28th of May is a Wednesday, not a Thursday.

What happened?

The date conversion routine that generates the DOW string does a bunch of odd stuff, but seems to work correctly; it certainly works in other parts of the code, and generates the right string there.


The UTC time seemed to be converted to local time twice, but that wasn’t the culprit; surprisingly, no-one is killed in an explosion of silicon splinters when that code is double-executed. Whatever.

Could it be that the system supplying you the time of that event in UTC is off by a year? One year into the future. That would give you that behaviour.

Check it.

Fixing the time

Ed Bott on how to fix the time sync in Windows (and the godawful error: “The time sample was rejected because: The peer’s stratum is less than the host’s stratum.”)

He notes this list of alternate time servers in the US. It makes sense for reliability purposes to choose a server close to you; indeed some big corporates run their own time servers.

As it turns out, Microsoft has a KB article documenting a number of servers around the world: 262680: A list of the Simple Network Time Protocol (SNTP) time servers that are available on the Internet.

Choose one near you today!

Firefox Spellcheck dictionary

A couple of times now I’ve had to hunt down the location of the dictionary for Firefox because in the popup for a misspelled word has ‘Add to dictionary’ too close to the word I want to change to (and now I’ve inserted a misspelling into my personal dictionary).

The location of the dictionary for Firefox (under Windows) is: somewhere under Documents and Settings is the file persdict.dat.

Maybe this time I’ll remember it. I suggest this behaviour shows a usability problem.

Wow, how did I miss the Mechanical Turk?

Amazon Mechanical Turk is an astonishing idea – an Artificial AI marketplace. Basically, there’s an API you can call to get humans to do tasks (oddly enough, they want to be paid). Currently, a big favourite for the tasks is transcribing podcasts. I can see that it would be a cheap way to truth a set of training data for AI systems, like number plate detection / recognition.

An artist has used the Mechanical Turk to acquire 10,000 hand drawn left-facing sheep and put them on a site for your viewing pleasure – plus, there was an exhibition of the collectable stamp sheets etc (you can buy the as stamp-sheets for only $20 a sheet). Given the images cost less than a cent each to acquire, he may be a bullshit artist.

The Turk is an example of what Wired calls Rise of Crowdsourcing – Remember outsourcing? Sending jobs to India and China is so 2003. The new pool of cheap labor: everyday people using their spare cycles to create content, solve problems, even do corporate R & D. It’s about the markets, people. These are markets for micro-transactions – micro in their repeatability, or micro in their value.

Sucky factorial calculators

Look for “factorial calculator” on Google and you’ll take a long time to find a factorial calculator that thinks that 100! doesn’t have an ‘e’ in it. If you’re going to write a dinky little app like that, be aware that there are limitations to it and tell people. I’m not going to link to any of them, they’re all naughty applications that shouldn’t be allowed out in the real world. But Dima Stopel’s large number factorial calculator isn’t afraid to give you all the digits.

OmniNerd – Articles: Beating Traffic

Brandon U. Hansen tries to figure out if he can get to work faster – beating traffic.

The world is full of traffic and people who hate it. This article analyzes a year of data to determine if minor tweaks to departure times can significantly impact commute length – or if it is all out of the driver’s control.

Well, duh. Of course it does. But who’s going to go to work at 11:00 and return home at 19:00?

Anyways, looking at his opening figures is weird, because he says that driving to work soaks up 100 hours a year, and involves 15,000 miles – which implies an average velocity of 150mph. Even if he forgot the trip home, you’re looking at 75mph (120kph), which is unlikely unless your cousin is a cop or you live right on top of an Autobahn. Perhaps they’re driving to work in a rocket tractor – which I always thought was one of these, with one of these attached – but it seems most of the net thinks they’re one of these, and that would never work!

Play School dumps clock with Big hand, Little hand

Monday morning I discovered that Play School has dumped the clock with Big hand, Little hand. I was running late to work, and while Owen was being breastfed Play School came on, and I overheard the presenter reading out the time. No more Big hand on this, Little hand on that.

I remember that some kids got analogue clocks quickly when we learnt in grade 3, and others just didn’t get it. My dad has always said that digital clocks aren’t an appropriate way to measure time, especially in a wrist watch or wall clock – because that’s not the kind of time you’re interested in – 19:37 is actually half past seven, or maybe a bit after half past – the precision is false/meaningless precision.

Oh, and in case I’ve been misleading you, the hands are now called Long hand and Short hand; clearer to the kids because the hour hand, whilst shorter is also stubbier, which could make one think it was larger. And no longer in this day and age is shorthand as common as it once was, so the risk of confusion there is reduced. And in a subsequent viewing of Play School (I’ve been home sick – stupid baby), they were back to Big and Little, so perhaps the presenter misspoke.

Oh well, time marches on…

While we’re here, a little tip: time should only ever be stored and (inter-system) communicated in UTC. If you’re designing a database that has time fields, for The Love of Sweet Merciful God store the time as UTC, regardless of the hoops involved in adjusting to that time zone. Because, when it comes to query, and you’ve got to start allowing for the fact that when daylight savings ends there are two 02:17s – and one came before the other; in addition you have the difficulty of knowing if the time on the clock was wound forward/backwards correctly – Australia has so many states with different time zones, and a propensity to diddle around with when the changeovers are meant to happen. Whilst effective, it is not desirable to use the “solution” one company I worked for went for – shutting all the machinery down for an hour when the clocks rolled back (especially given this was at the peak of processing for the day, not some idle time – and I mean business processing, not running the computer centre!). And figuring out when something happened in one time zone, compared to another, is a nightmare if all you’re storing is localtime; but equally, getting comprehensible output is tough if you’re not storing localtime alongside UTC. Here endeth the lesson.

Thursday night I had a dream – no, stick with me, this is relevant and interesting. Anyway, for whatever reason I had to build an analogue-to-digital clock converter, out of Lego. Now, I know how I did it in the dream, and on reflection, it would have worked. As a hint, I used Lego Mindstorms (in the real world I don’t earn enough to own, or for that matter to have ever touched, Mindstorms).

How would you construct an analogue-to-digital clock converter out of Lego?