tomsaffell

So many well informed comments here I have to ask... and please forgive the spam... If you're interested in this area and might be be interested in working at YouTube then please email me using the address in my profile. I work in this area at YouTube.

> I wonder if there's a converse effect?

If the inaccuracies are due to the meter's assumption around the current waveform then there probably is a way to exploit that. I suspect (from the little I have read) that the meters in question are: 1) assuming a sinusoidal current wave form, 2) assuming that the peak observed current is the peak of that sinusoidal wave, 3) effectively integrating under the area of the curve (with respect to voltage) to estimate total power. If all that's true, then to 'cheat' you'd need a device for which the current waveform looks like a 'fat' sine wave - same peak height, but wider peaks and steeper gradient through zero. In the extreme case it would be a square wave, but that might have too much harmonics.

For example, you could create a heater (a simple resistive load) that alters its resistance at 60hz. When the voltage reaches its peak the heater would have its 'normal' resistance, and during the next 1/4 cycle it would ramp down its resistance, reaching a resistive low as the voltage passes through zero, and then ramp up its resistance for the following 1/4 cycle until it reaches its 'normal' resistance at the peak 'negative' voltage (1/2 a cycle from where we started), then repeat over.

I think that would trick a meter that only looked at peak current and assumed a sinusoidal waveform. But who knows what other things it might break...

Yeah, terrible timing. A CNAME record conflicted with our new SSL cert - renewing the cert forced us off VIP on to SNI (per GAE's deprecation schedule for VIP) and the CNAME we had in place for VIP then prevented DNS from resolving properly - at least that's what I think happened ! (videopixie engineer :)

>repackage what actual news organizations publish with ideological slant.

Which is what The Economist does - it's been doing so since 1843. Current circulation is over one million, and it's widely respected, though not by all...

Good job you didn't have a third roommate who worked in printing...

Good to know! How do you handle Safari 8?

Hope this doesn't sound nitpicky, but I'm curious about the implication that 'Yelp customer service' would be considered 'middle class'. Were you specifically meaning that, or was it just co-incidence? I'm curious because I find that the issue of 'the hollowing out of the middle class' hinges on the definition of 'middle class', so I'd like to know how this might help inform the definition. thanks.

Looking at the 'before' and 'after' designs, I'd say that the biggest difference is that in the old design the first question you ask (via a textbox) is one to which the user already knows the answer ('what icon do you want?'). Whereas in the new design the first question is 'please pick one of these three options (with which you are not already familiar)'. So the questions requires the user to think. See Dont Make Me Think :) IANAUXD, just my $0.03

Interesting and worrying. Could you give more details on the "low status" and the "small shelf life"? Especially as it relates to gender (if indeed it does - forgive my ignorance).