* losing unsaved changes
* losing data due to a single point of storage failure (e.g. laptop harddisk crash / loss)
* savers remorse - losing previous versions of documents due to saving (losing state/"delta" information between versions)
* wasting time naming files, creating folders
* data access (single point of) (convenience)
* script failure; losing submitted information (e.g. Twitter XHR request)
1. minimize data loss. the probability of data loss should negligible, given the abundant availability of storage and ambient bandwidth.
2. minimize authoring overhead. content authoring and editing should have minimal UI overhead (naming, saving, filing, putting in folders).
3. (information openness)
* Gmail/GoogleDocs autosaving
* MS Entourage / MacOS 2000 - autosaves drafts of new messages
* MS Office autosave (once a minute)
* OpenOffice autosave of even untitled (never saved) documents
* S3 (Amazon)
* network layer versus service layer?
* Dojo
* Erasure Code http://en.wikipedia.org/wiki/Erasure_code
* Turbo Codes? http://en.wikipedia.org/wiki/Turbo_code
* persistent, broadband network connections
* Eliminate the notion of "Save". Everything is constantly persisted. There should be no data loss possible from sudden power outage, storage device removal, etc.
* incorporate backup into network-level protocols
* free backups by a company that openly mines data and shares the benefits with the public
--> what if we could guarantee .9999999% the integrity of your data BUT you have to make your information completely public
value in losing information?
privacy
User perspective: i dont care how it's implemented, but it shouldn't be possible to lose information.
Implementor's perspective: server-client communication?
inherent relationship to information openness
presence of versioning/past versions:
public by default, private is the exceptionPage Information
|
Wiki Information
|
Recent PBwiki Blog Posts |