Yay, some doom

[Menelaus]

The trouble is that you have to know that enough other people will subvert the system simultaneously to have confidence that you won’t be the one brave buy who gets let down by everyone else and caught out. Perhaps our resident mathematician Menelaus could develop a formula to determine precisely how many people you would need to commit to subversion to gain and spread this confidence?

That's an interesting problem; the economics piece of that would be the difficulty in ensuring the commitment - especially if the Jacq-booted one decides that she's going to start coming down heavy on it.

[EternalOptimist]

The trouble is that you have to know that enough other people will subvert the system simultaneously to have confidence that you won’t be the one brave buy who gets let down by everyone else and caught out. Perhaps our resident mathematician Menelaus could develop a formula to determine precisely how many people you would need to commit to subversion to gain and spread this confidence?

People like David Davis have already stuck their neck out on this. It wont take much to turn it into an avalanche, esp if HM opposition oppose it

Maybe Prince Charles will take a view on this

[Menelaus]

People like David Davis have already stuck their neck out on this. It wont take much to turn it into an avalanche, esp if HM opposition oppose it

Maybe Prince Charles will take a view on this

Once he's finished knocking-off horse face and talking to his trees? Erm, yeah. Does anyone think he'd be much of a warrior king?

[Mich the Tester]

People like David Davis have already stuck their neck out on this. It wont take much to turn it into an avalanche, esp if HM opposition oppose it

Maybe Prince Charles will take a view on this

As EO says, David Davies has stuck his neck out. I believe quite a few other senior opposition figures have done so, and the LibDems are listed as supporters of No2Id, as are some Labour MPs. Trouble is, Jacqboots has shown she has no qualms about letting the police arrest MPs for doing perfectly legal opposition work, so she probably won’t think twice about having David Davies, Menzies Campbell or Nick Clegg arrested. It might trigger some trouble though, if enough people can be persuaded to commit their support to those MPs who will take the risk.

[conned tractor]

I am the Stig of geekery.

Determination of good:bad is based on the dataset - for example, "good" might be (in the case of a medical statistics dataset) "cured" and bad would be "dead" - everything else would be an indeterminant.

Problems with this are obvious:

1. HUGE indeterminant data set meaning that separation statistically is made easier

2. Medical datasets are an easy and obvious good:bad selection issue

Kudos for the stig reference.

So, depends upon your dataset, in a sort of qualitative manner.

I am used to quantative data, usually sensor data using a linearised model + Kalman filter + simple statistical evaluation (SSE, etc) + heuristic determination of thresholds to determine faults ( with reasonable reliability). And from a practical approach - not a pure maths approach - although it would be useful to have more maths under my belt.

The question I asked was because I wondered if there would be a better approach to the determination of the thresholds for this type of system. Have also tried use of x*sigma deviation based threshold to put some theory to it but this requires analysis of the residual data to determine the threshold which I wanted to avoid.

Good luck with the research and apologies for hi-jacqi-smithing the thread.

[Menelaus]

As EO says, David Davies has stuck his neck out. I believe quite a few other senior opposition figures have done so, and the LibDems are listed as supporters of No2Id, as are some Labour MPs. Trouble is, Jacqboots has shown she has no qualms about letting the police arrest MPs for doing perfectly legal opposition work, so she probably won’t think twice about having David Davies, Menzies Campbell or Nick Clegg arrested. It might trigger some trouble though, if enough people can be persuaded to commit their support to those MPs who will take the risk.

Absolutely. And the right to protest has been curtailed in Parliament Square - indeed, I believe that Parliament created an Act recently specifically to get rid of their very own Banquos ghost, Brian Haw.

[Mich the Tester]

Once he's finished knocking-off horse face and talking to his trees? Erm, yeah. Does anyone think he'd be much of a warrior king?

Hmmm, I don’t know, but don’t underestimate the quiet guy. Remember Henry V;
In peace there's nothing so becomes a man
As modest stillness and humility;
But when the blast of war blows in our ears,
Then imitate the action of the tiger;
Stiffen the sinews, summon up the blood,
Disguise fair nature with hard-favor'd rage;
Then lend the eye a terrible aspect...

[Menelaus]

Kudos for the stig reference.

So, depends upon your dataset, in a sort of qualitative manner.

I am used to quantative data, usually sensor data using a linearised model + Kalman filter + simple statistical evaluation (SSE, etc) + heuristic determination of thresholds to determine faults ( with reasonable reliability). And from a practical approach - not a pure maths approach - although it would be useful to have more maths under my belt.

The question I asked was because I wondered if there would be a better approach to the determination of the thresholds for this type of system. Have also tried use of x*sigma deviation based threshold to put some theory to it but this requires analysis of the residual data to determine the threshold which I wanted to avoid.

Good luck with the research and apologies for hi-jacqi-smithing the thread.

You've got an interesting problem there. x*sigma deviation based thresholds would seem to be the way forward - if you can get hold of the residuals (is this possible for you?).

[Mich the Tester]

You've got an interesting problem there. x*sigma deviation based thresholds would seem to be the way forward - if you can get hold of the residuals (is this possible for you?).

Can you use all this clever stuff to answer my original question about how many people must commit to subversion?

[conned tractor]

You've got an interesting problem there. x*sigma deviation based thresholds would seem to be the way forward - if you can get hold of the residuals (is this possible for you?).

Yes, but its this I wanted to avoid. i.e. having to analyse the residuals first I am trying to find another way.