I don't see this issue as a search engine problem so much as it is a content format problem. The biggest problem with accessing "learning content" right now is that most of what we call learning content (SCORM packages) is closer to being a mini-application than it is to being a pure provider of content.
How can a search engine be expected to crawl a site that has its navigation locked away in a proprietary format....especially one that intentionally restricts the learner (and thus crawler) from proceeding until mastery is demonstrated. Most SCORM packages right now are monolithic SCOs that have their own proprietary navigation. How can a search engine be expected to crawl what is essentially a proprietary application. We don't expect a search engine to crawl the contents of a video (yet), we only expect it to index the public metadata that is associated with that video. While this model presents problems for search engine discovery, it has merit and utility derived from the educational experience it delivers.
"Crawl-ability" is related to a similar problem of "link-ability". Right now, there typically isn't a way to link to and deliver just a small, but relevant, part of content.
Obvious solutions to these problems include structured content formats or small SCOs tied together with Sequencing and Navigation. Both of these solutions have merit, but impose strict requirements and limitations that discourage broad use.
Increased use of metadata is another option that should be encouraged and explored...again though without imposing undue burden on content creators.
Is the best alternative to continue to allow for black box, "application-like" content, but also let such content provide alternative versions / navigation paths that expose the actual learning content in a crawl-able/link-able manner?