It would be really useful to get a definite idea on exactly how VO works. Rich and Alastair mentions that it is possibly a combination of the DOM and visual navigation which in this context we could refer to as the off screen model (OSM). But I would suggest that this may only be in the context of browsing the web. When using other screen readers to browse the web say JFW, this is done in a "virtual" mode using the off screen model. When the user comes across a form element that they need to enter some content they then enter "forms mode". This is in effect JFW actually interacting with the page itself and not with a "virtual" version or using the OSM. HAL interacts directly with the DOM AFAIK and does not use the OSM at all.
Also when a VO user is doing more OS level interaction, the mode must be different, does it still use the DOM or other different DOM types depending on the application being used? Josh
