Skip to content

Latest commit

 

History

History
38 lines (38 loc) · 1.91 KB

File metadata and controls

38 lines (38 loc) · 1.91 KB

The stack order

The algorithm doesn’t care

  • I have discussed how HETI calculates global unhappiness and tries to minimise it in the allocation process in a separate article
  • This is meant more as an endnote to that one
  • I will attempt to explain why the exact order of the stack makes no difference

Consensus reality

  • If we imagine our algorithm as a jet-powered bouncy ball bouncing through a landscape, the algorithm basically jumps and bounces over its hills and valleys until the rocket runs out of steam and falls into the deepest valley.
  • It calculates your unhappiness based on what position the hospital you got allocated to is in your preference list.
  • The natural consequence of this is that the only meaningful benefit of the stack is when used in combination with placing your favourite hospital first, as it essentially forces the system to reduce unhappiness by moving people into their first preferences
  • If you don’t get your first preference, what hospital you get instead essentially comes down to random chance. If everyone has the same preference list, there is, by and large, no particular advantage to swapping anyone from a non-first preference to a non-first preference.
  • Broadly speaking: *the only requirement for the stack to serve its functional purpose, is for there to be a consensus on exactly what the order of the stack is.*
  • And for that reason, any attempt to change the exact order of the stack, unless universally accepted, is frivolous. It may even be seditious.

Updates to come

Suggest more on Github or via PM/comment

  • [ ] Maths for nerds
  • [ ] Addressing proposals to change the stack
    • [ ] Hornsby vs Westmead
    • [ ] RNS vs RPA
  • [ ] Effect of consensus-breaking on placement
    • [ ] By proportion and number of variant stacks
  • [ ] Is the stack just a myth?