Inviability Risks

From RdC-Wiki
Jump to: navigation, search

Reasons why a web annotation system as proposed might not be viable.

Usability

  • (statement) People just won't use a complex system unless they are in real need
  • (why is it complex) People has to select the exact location for the annotation, and select the correct type of annotation.
    • (follow up) It might not be so difficult necesarily, but might need to split between the basic features (commenting with reduced options with a supportive interface), and advanced ones (e.g. preparing an advanced comment linking diferent locations, with concerns for copying a verbatim of sources).

Not a novelty

  • (statement:Bigger guys tried before) Bigger players with brighter minds have already walked these paths.
    • (objection) I might be lucky enough to have a better approach, or to be at the right place at the right time. But most probably not.
    • (follow up) Lets try to find out or figure out what these people tried, and why they didn't make it.
    • (follow up:Ellusive,AI) some problems (such as AI) seem to be ellusive. Maybe too many concerns that must be attended at the same time (too big to chew), or maybe because there are issues (like knowing step by step how a human mind would do in a wikipedia:Chinese room).

No clear advantage

  • (statement) People won't use it if it doesn't give them a groundbreaking advange
    • (concretely) People use wikis and forums for their purposes. That might not be perfect, but is good enough for most cases.
      • (concretely) Forum controversies might end up being bulky and messy, but people dig into it. If we just offer a way to clean it up, they won't bother.
        • (however) If we offer a way to cleanly best lesser adversaries, and graciously evade enquiries that don't deserve being responded, then we will be getting their attention, as they spend lots of hours in fruitless discussions.

Too big to chew

  • (statement) Even simple things take longer than expected to be done. Complex things take long even figuring out your way through.
  • (simple is done) We might implement for example an open annotation system without any concerns, but Annotea or even RDF-databases are up to that.
  • (complex is too much) A system open to the public must pay attention to many concerns
    • (suggestion) Maybe then we should extend other systems which already pay attention to those concerns

Critical Mass

  • (statement) Some systems won't be used if no one uses them already. Therefore as an start and final point, nobody uses them.
    • (for example) if some one designs a better e-mail system, or a better auction system (better than ebay), people won't be interested unless they can mail their colleagues and find buyers or stuff. People won't be using a redundant system (which will be only useful to reach some people, who will be using the previous system anyway) unless there is enough people using it to pay out.
    • (controlled group scenarios) Such novel systems work better if the audience is in a controlled group (a research group, a company) who are more capable to switch at the same time.

Abuse

  • (statement) Like in Murphy's law, if using a system in a wrong way brings benefit or satisfaction, then it will become the right way to use it.
  • (concretely) There are activists (politics, religion, etc) that will use any loudspeaker they find at hand, and will game any system to give themselves dominance. An annotation system might easily be subject to edits and moderations wars.
  • (concretely) There are copyrighted resources (like music or movies). It might easily happen that the references to copyrighted material might be annotated, directly and indirectly with P2P download information, which the copyright holders will frown upon and litigate.
  • (concretely) There is spam, scams, and illegal activities
  • (concretely) People might blame and sue you for allowing some annotations, and for deleting the very same annotations

Proposals:

  • (smaller audience) Won't be much an issue as long as the user audience is reduced
  • (enforce registration) Requiring registration is the first and foremost measure to control "irresponsible" publications
  • (report button) A "report"/"panic" button allows administrators to focus on bad stuff and users, and remove them.
  • (captchas) Captchas might stop noxious scripts, but they are defeatable, specially if there might be a strong interest.
  • (filters) Some sites have filters to help keep the bad stuff off (e.g. YouTube might detect if you upload a copyrighted video clip)
  • (microfee) A small fee per transaction might be an ultimate barrier for "volume infractors", as well to give a good pay-per-use monetization. Unfortunately it will keep legitimate users off, and the management of the micro-money might become a bigger issue than the service itself.

Scalability

  • (reference system) Systems like Wikipedia or slashdot need clusters of servers to keep up with their legion of users.
  • (difference) If we are talking about a system which is not only accessed when it is directly visited, but when any other site is visited (even if just looking for annotations availability), then we might have a huge scalability problem (maybe in the order of magnitude of Google, or even worse).

Proposals:

  • (HTTP cache) If requests are regular HTTP request, they might be cached
    • (however) customized responses wont benefit
    • (however) cache sysops might frown upon that annotation service which consumes so much (which is debatable, but some people just need an excuse to try squeezing some more money, or to blame others for their problems).
  • (Premium/basic duality) Just like direct download sites, we might like better out paying patrons than the freaking freeloaders trial users.

Who pays the bills

  • It would be nice to have a service free for everyone (like the Wikipedia): The more contributors and readers the better (#Critial_mass).
  • Some systems use advertisements
    • However some sites (like WikiMedia) avoid them for several reasons (sensible adds might even be an added value, but it still is a random input in your content).
    • Some systems (like an annotation server) are used not (only) by humans, but through program APIs. Programs are not known for succumbing to publicity. Even if there is not an available API, some programs might parse the human presentation output and get the data. Some systems like Google Translate stopped offering an API after being leeched by third party software.
  • Some systems depend on donations (WikiMedia Foundation)
  • Some systems are restricted to a userbase (which in the worst case at least limits the hosting bill ammount)

Resource mutability

Annotations are made over contents, but contents make change overtime. We might need a way to keep track that the linked resource is still what it is supposed to be. However, since we don't control third party resources, that is only achievable to some degree.

  • (optimistic scenario) We might expect the content editors to preserve the content, and if ever, only do minor changes.
  • (scenario) However a site might be hacked, and its contents defaced with spam or whatever.
  • (scenario) A site owner might want to do a clean up of old resources regardless possible annotations.
  • (scenario) A page might not be the proper location for a content (e.g. a blog front page which posts shift down and dissapear as more contents are published).
  • (scenario) Misschievous people might change the contents of a page just to play other people. Just to name the most "straight-forward" example: Attract links/categorization with a kitties photo gallery, then change those with other kind of "pussy", and then blame the link owners for referencing pornography.

Proposals:

  • A hash of the content might be stored to check for content integrity.
    • (however) Hashes up to MD5 are compromised and might be circunvect. Not only a evil publisher can malevously alter her contents, but also introduces a new scenario where a rogue annotator blames the original publisher for some content that happens to have the same hash.
  • A verbatim copy of the content might be stored to avoid losing track of the original.
    • (however) some materials might be copyrighted. Even private copies might be disallowed.
      • (however) typical internet forums, or divulgation sites do not have such constraints.
  • A conscious publisher might interact with the available annotations, in this case notifying of the modification, so the system and the annotation owners can check and correct them. The publisher might even introduce link anchors and permanent URIs to the contents, or support some version control over her contents.