Google has the "same content" policy in place to prevent cloaking:
Does this scheme open the door to cloaking?
Cloaking is serving different content to crawlers than to users in response to a given URL. This is generally done with the intent of boosting one's ranking in search results. Cloaking has always been (and will always be) an important issue for search engines, and it's important to note that making AJAX applications crawlable is by no means an invitation to make cloaking easier. For this reason, the HTML snapshot must contain the same content as the end user would see in a browser. If this is not the case, it may be considered cloaking. See our article on cloaking for more details.
Google cares most about including text content and links that are not shown to the user. That is a big no-no that can get your site thrown out of the Google index. Using a pared down markup with the same text for HTML snapshots passes my sniff test. It isn't deceptive. It isn't designed to make your site rank higher. I would be inclined to implement it that way if it makes the development easier.
You would want to include all the tags that Google might possibly care about for ranking purposes. Links (
<a>) would be especially important to include just as shown to the user. I would also include images (
<img>), headers (
<h3>, etc) as well as block level tags that break content into sections.
From a change managament perspective, you also want to make sure that any time you make a change for users, you also make the corresponding change to the HTML snapshots. Using a method that produces pared down HTML implies that you are not actually generating the snapshots from the UI and that there is a separate development for it. That sounds like a problem waiting to happen to me.