WEB GOOD:
- people put data there
- standards
- simple, easy to "publish" hypertext
- cheap(ish)
- empowerment of individual publication
- accessible
- easy to browse
- low-end users (***)
- lots of $$ driving it (***)
- cheap to use
- inline graphics (in text)
- equality of relevance to all data
- enables sharing
- searchable
- keyword search
- distributed search
- big user base, link/access patterns can drive metrics
- fuzzy queries possible
- software distribution
- apps (and upgrades to them)
- applets
- uniform naming of network resources
- standards
- document format (HTML)
- standard application (part of OS)
- services/commerce/information (w/direct access)
- shopping, travel, stock trading
- fresh, updated 24x7
- at your convenience, on your own time
- can get things at good price (ex: Yahoo shopping experience)
- distributed community-building
- customization of information services
- custom pull
- custom presentation
WEB BAD:
- unpredicatable performance ("world wide wait")
- no quality control: high volume, high variance in (quality) utility
of information
- lack of classification
- publishing faster than classification
- security retrofitted after the fact, not good
- personal info often transmitted electronically, more on public
channels
- lack of privacy (behavior and info)
- naming structure has (apparent) semantic meaning - semantic ids
- breakable / overly flexible document standards (HTML)
- "eyeball encoding" of info
- for humans, not for machines
- underutilized networked resources
- hard to find what you want (***)
- different ways to get similar results
- hard to compose right query (keywords)
- hard to know what the data set is (time-varying)
- hard to know what the queries mean
- hard to debug (no explain facility)
- hard to judge completeness of query results (GIGO)
- query bait (deceptive metadata)
- lack of non-text search facility
- some image search
- mostly exact (word-wise)
- no structured search
- poorly maintained data
- evil HTML hacks ("trap"-like sites)
- long transactions are natural
- time outs
- no guarantees on ACID transaction
- no end-to-end confirmation (general problem)
- bad product evaluation/inspection capabilities
- socio-political issues
- huge, important, but out of scope
- replication/caching/multicast and other issues not auto-solved
- spiky traffic
- lack of human organization of data
- fixable at some scale
- but: different classifications, fast rate
- lack of good natural language queries (general problem)
- software quality
- decentralization - ex: SW airlines