Powered by glolg
Display Preferences Most Recent Entries Chatterbox Blog Links Site Statistics Category Tags About Me, Myself and Gilbert XML RSS Feed
Thursday, Jan 29, 2015 - 21:02 SGT
Posted By: Gilbert

An Opportune Return

Mr. Ham: *peeks into cubicle yet again* Hi, Mr. Ro... that's strange, I could have sworn that I had seen him come in today. Now, where's that hamster?

*muffled voice from under desk* He's not in. Please go away.

Mr. Ham: That's a right shame. But just as well. Joe? Just leave these stores of highly unstable explosives in this here empty cubicle for the time being, wouldja?

Mr. Robo: *crawling out from behind wastepaper basket* Fine, fine, you got me. And no, I wasn't shirking my duties down there. Also, I won't be taking any more breaks. Nor subsidize any more of your dinners. So, could you just leave me alone and not drag me into any of your affairs?

Mr. Ham: Oh, I was merely dropping by to inform you that the human's active again. You're off the hook for now.

Mr. Robo: Why didn't you say so earlier?!

Me: It's okay, I'll take over from here.


What Can You Say To This?

"...Singapore must avoid a situation where government subsidies and policies turn into an auction in the elections."

- new labour chief and possible future PM, 2015


"...The answer is that there has to be a distinction. Because the PAP wards supported the Government and the policies which delivered these good things..."

- incumbent PM explaining why not supporting them in elections rightly means that your subsidies are delayed indefinitely (but hor, this is not an auction!), 2011


January's usually a good month for labour woes to make an appearance, and right smack on cue, The State's Times has given voice to eateries turning away diners and vacancies rising as locals spurn low-paying jobs. Productivity's still sinking, though, but we'll leave that for later. Observe first the helpful table that made it to the front page of Wednesday's State's Times:

Five occupations with the most vacancies in 2014
OccupationVacanciesMedian gross wage% unfilled for six months
Shop sales assistants3730S$174148.6
Security guards2250S$167863.0
Waiters2200S$135265.2
Cleaners1850S$100066.0
Teaching professionals1770S$77451.5


Ignoring the last row, one could imagine that the situation's not that bad - we could simply let in ten thousand foreign workers, and our sales assistant/security guard/waiter/cleaner situation would be completely resolved! Well, we added something like thirty thousand last year despite the "big slowdown" in any case, so why not allow in a few more to resolve the issue once and for all?

Well, as noted here slightly more than a year ago (and not, insofar as I have seen, recognized elsewhere), this wouldn't eliminate - or probably even significantly reduce vacancies - since the newly imported workers would themselves drive new demand! Or, bring in four thousand shop assistants, and you'd soon see more shops opening to cater to them, who would then want - guess what - more assistants (and at low wages)

Definitely, if you were seeking to raise absolute GDP numbers, then sure, this is a nice property to rely on. Shame about the productivity and crowding, but sacrifices have to be made.

Or, we could interpret this in terms of supply and demand. What the above State's Times-approved stats are showing is that, when employers are offering S$1000/month on average, they are not able to get as many cleaners as they'd like.

What the mainstream media seldom points out, is that there is actually absolutely nothing wrong about this!

An example: as the table points out, at a S$1000 median wage, there are 1850 vacancies for cleaners. What would happen if the S$1000 became S$1250? Well, I'd suppose the number of vacancies would fall. And it it became S$800? One would expect the number of vacancies to rise, as some cleaners quit (or not, in the extreme case of a countrywide recession that forces people to take on second or third jobs)

So the question: what makes the tuple of {1850 vacancies;S$1000} so special? It could as easily have been {1200;S$1250}, or {2500;S$800}, or any one of any number of such combinations. Furthermore, one could suspect that a good portion of the vacancies would be due to employers offering below the median wage - but if Kiam Siap Pte Ltd. is not getting any applicants for cleaning staff at S$600/month, is this in any way meaningful? I don't see how we'd ever eliminate vacancies with subcontrators lowballing, anyway.

One oft-repeated objection of the mainstream echo chamber is that there is a fundamental mismatch, and that raising wages would not cause these vacancies to be filled (?!). As far as I could discern, and supported by the response to S$3000 dishwasher jobs (which was then hurriedly revealed to involve near-illegal twelve-hour days), this is more or less bullshit.

I'll leave the political analysis for next time.


The Triumph Of Stats (Or: How Big Is Your D***?)

With all the fancy buzzwords about the bigness of data being flung about, it is often forgotten that rather more intuitive observations can frequently be drawn from concise snapshots. Take, for instance, the 29.4% arrears rate for the AHPE town council that was lambasted by an incumbent minister a couple of months back, which was noted here to be plainly incongruous given aggregate resident behaviour.

Well, the mystery has been resolved - the true arrears rate for 2013 was actually only about 5 to 7 percent, far more in keeping with other town councils. As to why there was such a huge error, it was attributed to manual double-counting, which was due to the previous computerized system being withdrawn, which brings us full circle to AIMgate, where we find ourselves up against a completely legal dead end.

So that's that.

Definitely, it would have been better had one of the guys working on the reports recognized that 29.4% was totally ridiculous, or had some incumbent representative expressed doubt and concern about why the citizens were in these dire straits, rather than immediately going for political point-scoring. Neither did our award-winning mainstream media think to investigate, when one of their trademark polls - which would for once be appropriate - could have quickly revealed that it was likely that the true arrears rate was far lower.

But eh, this makes the opposition look bad, so okay lah. That said, the SDP Sec Gen seems to have rubbed someone (this time, a new recruit) the wrong way yet again, though his dismissal might not be totally without reason. Whatever the actual truth behind it, their road remains long and hard.

We move now to the States, where the story of the week was the forecasted (near-)record blizzard on the East Coast... which turned out to be bad, but not that historically bad.

Now, one can't exactly blame the forecasters for doing the responsible thing - imagine if they had held back on their best judgment, only for unwary commuters to be stranded - but it transpires that the Weather Channel got it spot on. As Slate explains, this was due to them using more separate models and weighing them more equally, as opposed to their competitors' reliance on just a couple of the best ones.

While this is probably too little to draw any lasting conclusions from - particularly since it is unclear how much of a meta-model those "best" ones were - it is perhaps worth a note that this is a heads up to the power of ensembling (and crowds): collate enough independent and reasonable opinions, and their faults tend to cancel each other out.

And, just before that, the talk of the town was... balls. In the AFC championship game, of which the winners would proceed on to the Super Bowl (i.e. America's one true holy day), an intercepted ball was handed by a member of the defending team to his equipment staff to keep as a souvenir, only for them to suspect that the ball had been underinflated.


It probably wasn't *this* bad
(Source: stcatharinesstandard.ca)


Now, you might be scratching your head. Isn't this what happens to balls? Lose their air gradually? That's what pumps are for, right? And also, aren't both teams playing with the same ball anyway, so what's the big deal?

On that first, it is well-established that sports teams much prefer to play with their own balls (out of familiarity, I reckon) - recall the arrangement worked out for the 1930 soccer World Cup. And on the second point, unlike real football (don't hit me, Madden fans), nope - effectively, only one team - the attacking one - ever actually handles the ball in practice. The exception, an interception return, is so rare as to be effectively negligible.

One could wonder why in the world would the NFL not simply supply the official balls (one thing FIFA has gotten right), rather than having their referees waste time inspecting them, and it seems that it was due to the demand from star quarterbacks. Fair enough, since they'll be the ones doing most of the handling... so why not, as long as they can be trusted not to tamper with the equipment. Right? Right?

Of course, such tampering is almost a tradition in American sport - if some advantage, no matter how small, can be derived from fixing up the goods, you can bet your ass that it has been tried already. Baseball has seen spitballs and corked bats, and the NFL themselves had apparently forgotten what happened when they neglected to set a limit on how much adhesive players could apply to themselves; quite predictably, this soon became "as much as is humanly possible".

But anyway, back to the stats. The New England Patriots started off by denying everything, as expected, while suggesting various not-implausible reasons why their balls might not be as big as specified. Enter the data scientists.

By scouring past statistics, analysts soon produced a damning indictment - the Patriots' rate of fumbling was so low, compared to all other NFL teams, as to be nearly off the charts. Fitting it to a normal distribution, one would estimate the chances of such an outlier happening to be on the order of 0.006%! But, couldn't it be that they simply are that good?

On to part two. Comparing the fumble rate from 2000 to 2006 (before teams could supply their own balls on the attack) against 2007 onwards, they found that the Patriots improved from a decidedly-average 42, to an out-of-the-world 74.

Bearing in mind that none of this is rock-solid evidence - the Patriots could simply have improved drastically, or their offensive unit could have been really, really conditioned to their own still-within-regulation balls, with the deflations truly a honest mistake - the data does point to shenanigans. As the NFL are trundling out the storied physics department of Columbia University (more on them next time), we should have a more authoritative opinion any day now.

So we come to the moral of the story, according to Mr. Ham:
  1. Don't get caught
  2. Have big balls
  3. Fix those bloody beancounting meddlers





comments (0) - email - share - print - direct link
trackbacks (0) - trackback url


Next: A Game Of Half-Truths


Related Posts:
A Life Quantified
The Numbers Game
Burps And Bingo
Anatomy Of A Kick
In Another Class (Part Two)
Back to top

Copyright © 2006-2025 GLYS. All Rights Reserved.