Author Topic: The Algorithm Debate  (Read 1077 times)

Offline Omollo

  • Moderator
  • Enigma
  • *
  • Posts: 7143
  • Reputation: 13780
  • http://www.omollosview.com
    • Omollosview
The Algorithm Debate
« on: August 21, 2017, 11:10:36 AM »
... [the ICC case] will be tried in Europe, where due procedure and expertise prevail.; ... Second-guessing Ocampo and fantasizing ..has obviously become a national pastime.- NattyDread

Offline Omollo

  • Moderator
  • Enigma
  • *
  • Posts: 7143
  • Reputation: 13780
  • http://www.omollosview.com
    • Omollosview
Re: The Algorithm Debate
« Reply #1 on: August 21, 2017, 11:10:59 AM »
... [the ICC case] will be tried in Europe, where due procedure and expertise prevail.; ... Second-guessing Ocampo and fantasizing ..has obviously become a national pastime.- NattyDread

Offline Omollo

  • Moderator
  • Enigma
  • *
  • Posts: 7143
  • Reputation: 13780
  • http://www.omollosview.com
    • Omollosview
Re: The Algorithm Debate
« Reply #2 on: August 21, 2017, 11:12:09 AM »
... [the ICC case] will be tried in Europe, where due procedure and expertise prevail.; ... Second-guessing Ocampo and fantasizing ..has obviously become a national pastime.- NattyDread

Offline RV Pundit

  • Moderator
  • Enigma
  • *
  • Posts: 37009
  • Reputation: 1074446
Re: The Algorithm Debate
« Reply #3 on: August 21, 2017, 11:54:27 AM »
Not seeing anything. Are you missing some url?

Algorithm theory: In situation where 41,000 polling stations are rapidly and progressively releasing per-polling stations results - this provides a near perfect uniform random distribution that should reflect the final results at every given time. It doesn't matter if 1K are in or 30k are in - the final tally will be the same - because this is near perfect random sampling - there will small changes here and there because maybe NASA or Jubilee strongholds have slightly better connectivity or less votes per polling station making counting faster - so they are likely to marginally report their results earlier or later - but overall as the sample grows - the marginal difference disappears.

For example.
1)If good random sample of 41K - say the first 1000 polling stations randomly transmit their results from every corner - it should represent the final tally - 54% verus 45%.
2)If good random sample of 41k - say the next 10K polling station randomly transmit their results from every corner - it should represent the final tally  -54% versus 45%
3)If whole set of 41k station all submit that information at ONCE - it should reflect 54% versus 45%.

In short this actually the basis of opinion polling - and this results vindicate IEBC.

You can compare and contrast with 2007 which was acknowledge by many to have been rigged for wild swings - Kibaki was withholding results from his strongholds - so he can message the figures before releasing them - so Raila takes huge lead - before Kibaki takes the lead!

There was no algorithm except
1) IEBC made sure each polling station had a max of 700 votes.
2) IEBC made sure each polling station would count and immediately transmit their results (by key-in)  - giving no avenue for anyone to message the numbers
3) IEBC made sure each polling station would later second a scan copy of Form 34A.
4) IEBC made sure Form 43A and polling materials would be taken to Const Ro - where counting and tallying again would be done - FORM 34C.

Under such circumstances we just needed the first 1000 polling station to know who'd have won. It an exit poll or opinion poll.