Upon further review I noticed some estimates that should probably be 
revised down:

On Mon, 27 Oct 2008, Ian Hickson wrote:
  

2. Interaction events

Scope: defining the relationship between user interaction (clicks, mouse 
movements, key presses, etc) and DOM2 Events, hit testing, and documenting 
any special behavior relating to HTML features.

Skills and knowledge needed: JS, DOM, HTML, reverse-engineering browsers. 
Access to multiple browsers and platforms necessary.

Estimated workload:
 - 6 months at 40h/week researching, reverse engineering and specifying
 - 4 months at 20h/week responding to immediate feedback
 - 12 months at 5h/week responding to further feedback
 - 24 months at 1h/week responding to further feedback
    

6 months for the initial research is probably longer than necessary. Upon 
further thought I think 2 months is probably enough there, provided the 
next 4 months do include further research in response to feedback:

  - 2 months at 40h/week researching, reverse engineering and specifying
  - 4 months at 20h/week researching and responding to immediate feedback
  - 12 months at 5h/week responding to further feedback
  - 24 months at 1h/week responding to further feedback


  

9. 2D Canvas

Scope: correcting errors and adding new features to the 2D graphics API.

Splitting this out would probably slow down HTML5 itself by a few weeks.

Skills and knowledge needed: 2D graphics, JS, DOM, HTML (in particular, 
experience with the 2D canvas API), API design, reverse-engineering 
browsers. Access to multiple browsers and platforms necessary. Ability to 
tactfully say "no" a lot a definite plus.

Estimated workload:
 - 24 months at 20h/week researching, reverse engineering, designing, 
   specifying, and responding to feedback
 - 24 months at 2h/week responding to further feedback
    

The above estimate assumes significant further work on canvas. If the 
editor is able to say "no" to requests, which I think in this case is 
highly desireable, then I would probably revise the above to:

  - 6 months at 20h/week researching, reverse engineering, designing,
    specifying, and responding to feedback
  - 12 months at 10h/week responding to feedback
  - 24 months at 2h/week responding to further feedback


The other estimates seem about right.