5 ms·

This is exactly what i was thinking, even though the finish time of Eastenders might not be what the BBC says it is, its not like it would be hard to write a pr

This is exactly what i was thinking, even though the finish time of Eastenders might not be what the BBC says it is, its not like it would be hard to write a program to listen for the "Dun, dun dundundunduuun" and turn up the power.

polymatter 14y ago
That's probably too brittle a solution. Eastenders ads, other programs etc. could conceivably play enough of the Eastenders theme tune to trigger unnecessary power surges.
- TamDenholm 14y ago
  Well it was a half sarcastic comment but theoretically you'd only have to have it active when the program is live, i'm sure theres some kind of computer system at the BBC that tells you whats on air at the time even if its unscheduled.
  - kintamanimatt 14y ago
    There probably isn't and the UK has more than just the BBC!
enneff 14y ago
I don't know. I haven't heard of any major outages as a result of this guy not doing his job properly, and if it's just the one guy... why replace him with software? It doesn't seem worth it.
- TamDenholm 14y ago
  Well i'm a typical software geek, if i see something that can be automated, i cant help but think of a way to do it, not saying that everything should be, but it almost always crosses my mind.
  - cperciva 14y ago
    Given the level of care required in writing software for running the national electrical grid, I think it's quite likely that writing the automation software would be more expensive than paying a human to deal with it. (Especially since this is just a small part of his day -- he presumably has other responsibilities as well.)
    - scrumper 14y ago
      Exactly. And you'd still need the man to watch the computer (we have unions in the UK too), and then of course the dog to bite the man. If it ain't broke...
    - breck 14y ago
      But when you automate things the benefits are often orders of magnitude larger than simply the cost savings from the jobs you replace.
      - chris_wot 14y ago
        Emphasis should be on often. I don't think a cost benefit analysis would show automation in this instance to be worthwhile. But it would be interesting to see if I'm right!
        lewispollard 14y ago
        I would assume that the National Grid have done such an analysis and determined that it isn't worthwhile, seeing as how... they still employ a bloke to do the job!
        chris_wot 14y ago
        I'm pretty cynical... I suspect the analysis was - do we allow the guy on duty watch East Enders and do his job after hours, and pay him a bit of overtime, or do we pend a lot of money for IT to come up with an automated solution that might not work th first time...
kintamanimatt 14y ago
The good thing about humans is they have very well-written exception handlers built in! Such a program is likely to be brittle and could fail unexpectedly requiring human intervention. There are also many other events that couldn't be triggered, such as the beginning of commercials for non-BBC shows, the beginning and end of popular sporting or other TV events that would trigger the same phenomenon. It's probably far simpler and cheaper to have someone sit there and press buttons, with the added bonus of high accuracy.
- muyuu 14y ago
  I bet the operator can also oversee that the power-up operation went well and can react to a number of scenarios, and also improvise. Given the importance of the operation I don't think having a person dedicate a little of his or her working day to that is unreasonable.
saidajigumi 14y ago
Fine, that solves the problem for exactly one ephemeral correlate of power consumption surges. What about the next one? If we're talking armchair first-stabs at automation, I'd much rather have 1) a large set of historical data from distributed consumption data feeds across the service area and 2) analysis of that same data live and with near-realtime latency. Given that, it seems plausible that a well-chosen machine learning based system could provide useful automated assistance. That said, there are a bunch of practical reasons why it probably never makes sense to take human judgement out of the equation, esp. given the potential infrastructure impact here.
politician 14y ago
An explicit signaling solution seems more appropriate here. Suppose the BBC were to embed a frame containing a time-remaining flag at various points in the broadcast. Then the power guys could write some software to watch the stream for these frames, decode them, and estimate the spike. If necessary, legislate the presence of the frame.
- ditoa 14y ago
  We do. They are called Cue Marks or Cue Dots. http://en.wikipedia.org/wiki/Cue_mark#Cue_dots_in_television http://en.wikipedia.org/wiki/Cue_mark#Cue_dots_in_television
  - danellis 14y ago
    Or something like Programme Delivery Control: http://en.wikipedia.org/wiki/Programme_Delivery_Control http://en.wikipedia.org/wiki/Programme_Delivery_Control
- jacquesm 14y ago
  What could possibly go wrong? That's why we don't let politicians design technical systems. They're bound to come up with solutions that sound good but that will not work in practice. Some problems with your approach: - partial loss of signal -> false positive -> power surge - loss of signal -> false negative -> grid overload - how do you signal the need for a little bit more power? And so on. The British solution is typical: it's simple, and it 'just works', the speed with which hydro comes online it is just not enough to be reactive but even a little bit of forewarning and they'll be fine. Heck if the TV fails he can call home to mum, or some other place. Solution of last resort: a slight brown out due to lack of information, but as long as this system has been in place you'd think if that were a statistically frequent thing that it would have happened by now. Compare it with the auto-traffic-info system present in most car radios. http://en.wikipedia.org/wiki/Radio_Data_System http://en.wikipedia.org/wiki/Radio_Data_System The alternative frequencies system works along the lines you suggest (though technically it's not a 'frame' but an entire channel with its own carrier wave). This fails so frequently that I usually disable it. Either you get switched to some totally unrelated station (without traffic info to boot) and then never switched back or you're listening to a CD or so and then you'll suddenly be in the midst of some radio broadcast and again, you're never switched back. Easy to propose, using an unreliable medium such as a TV or radio channel apparently very hard to get 100% right. And if you're going to change major parameters in the electricity generation capacity of a nationwide grid I assume you'd put the bar for reliability a bit higher than I do for my car radio.
toyg 14y ago
And then EastEnders changes the tune ever-so-slightly (without notifying National Grid operators, of course), and the program breaks. Or the soap gets moved, or cancelled, and the program breaks.

polymatter 14y ago

That's probably too brittle a solution. Eastenders ads, other programs etc. could conceivably play enough of the Eastenders theme tune to trigger unnecessary power surges.

TamDenholm 14y ago
Well it was a half sarcastic comment but theoretically you'd only have to have it active when the program is live, i'm sure theres some kind of computer system at the BBC that tells you whats on air at the time even if its unscheduled.
- kintamanimatt 14y ago
  There probably isn't and the UK has more than just the BBC!

enneff 14y ago

I don't know. I haven't heard of any major outages as a result of this guy not doing his job properly, and if it's just the one guy... why replace him with software? It doesn't seem worth it.

TamDenholm 14y ago
Well i'm a typical software geek, if i see something that can be automated, i cant help but think of a way to do it, not saying that everything should be, but it almost always crosses my mind.
- cperciva 14y ago
  Given the level of care required in writing software for running the national electrical grid, I think it's quite likely that writing the automation software would be more expensive than paying a human to deal with it. (Especially since this is just a small part of his day -- he presumably has other responsibilities as well.)
  - scrumper 14y ago
    Exactly. And you'd still need the man to watch the computer (we have unions in the UK too), and then of course the dog to bite the man. If it ain't broke...
  - breck 14y ago
    But when you automate things the benefits are often orders of magnitude larger than simply the cost savings from the jobs you replace.
    - chris_wot 14y ago
      Emphasis should be on often. I don't think a cost benefit analysis would show automation in this instance to be worthwhile. But it would be interesting to see if I'm right!
      - lewispollard 14y ago
        I would assume that the National Grid have done such an analysis and determined that it isn't worthwhile, seeing as how... they still employ a bloke to do the job!
        chris_wot 14y ago
        I'm pretty cynical... I suspect the analysis was - do we allow the guy on duty watch East Enders and do his job after hours, and pay him a bit of overtime, or do we pend a lot of money for IT to come up with an automated solution that might not work th first time...

kintamanimatt 14y ago

The good thing about humans is they have very well-written exception handlers built in! Such a program is likely to be brittle and could fail unexpectedly requiring human intervention. There are also many other events that couldn't be triggered, such as the beginning of commercials for non-BBC shows, the beginning and end of popular sporting or other TV events that would trigger the same phenomenon. It's probably far simpler and cheaper to have someone sit there and press buttons, with the added bonus of high accuracy.

muyuu 14y ago
I bet the operator can also oversee that the power-up operation went well and can react to a number of scenarios, and also improvise. Given the importance of the operation I don't think having a person dedicate a little of his or her working day to that is unreasonable.

saidajigumi 14y ago

Fine, that solves the problem for exactly one ephemeral correlate of power consumption surges. What about the next one? If we're talking armchair first-stabs at automation, I'd much rather have 1) a large set of historical data from distributed consumption data feeds across the service area and 2) analysis of that same data live and with near-realtime latency. Given that, it seems plausible that a well-chosen machine learning based system could provide useful automated assistance. That said, there are a bunch of practical reasons why it probably never makes sense to take human judgement out of the equation, esp. given the potential infrastructure impact here.

politician 14y ago

An explicit signaling solution seems more appropriate here. Suppose the BBC were to embed a frame containing a time-remaining flag at various points in the broadcast. Then the power guys could write some software to watch the stream for these frames, decode them, and estimate the spike. If necessary, legislate the presence of the frame.

ditoa 14y ago
We do. They are called Cue Marks or Cue Dots. http://en.wikipedia.org/wiki/Cue_mark#Cue_dots_in_television http://en.wikipedia.org/wiki/Cue_mark#Cue_dots_in_television
- danellis 14y ago
  Or something like Programme Delivery Control: http://en.wikipedia.org/wiki/Programme_Delivery_Control http://en.wikipedia.org/wiki/Programme_Delivery_Control
jacquesm 14y ago
What could possibly go wrong? That's why we don't let politicians design technical systems. They're bound to come up with solutions that sound good but that will not work in practice. Some problems with your approach: - partial loss of signal -> false positive -> power surge - loss of signal -> false negative -> grid overload - how do you signal the need for a little bit more power? And so on. The British solution is typical: it's simple, and it 'just works', the speed with which hydro comes online it is just not enough to be reactive but even a little bit of forewarning and they'll be fine. Heck if the TV fails he can call home to mum, or some other place. Solution of last resort: a slight brown out due to lack of information, but as long as this system has been in place you'd think if that were a statistically frequent thing that it would have happened by now. Compare it with the auto-traffic-info system present in most car radios. http://en.wikipedia.org/wiki/Radio_Data_System http://en.wikipedia.org/wiki/Radio_Data_System The alternative frequencies system works along the lines you suggest (though technically it's not a 'frame' but an entire channel with its own carrier wave). This fails so frequently that I usually disable it. Either you get switched to some totally unrelated station (without traffic info to boot) and then never switched back or you're listening to a CD or so and then you'll suddenly be in the midst of some radio broadcast and again, you're never switched back. Easy to propose, using an unreliable medium such as a TV or radio channel apparently very hard to get 100% right. And if you're going to change major parameters in the electricity generation capacity of a nationwide grid I assume you'd put the bar for reliability a bit higher than I do for my car radio.

toyg 14y ago

And then EastEnders changes the tune ever-so-slightly (without notifying National Grid operators, of course), and the program breaks. Or the soap gets moved, or cancelled, and the program breaks.