Theres many different types of web sites. Some new ideas are proposed below.
By using digg technology and combining that with the millions of news articles on google news the most popular news across every categorie can be shown to users. Users could then change there prefernces to include more of the categories they like, and it would show them more articles that people have "dugg" on those categories.
It may even be possible that the computers digg for articles themselves, but there vote is only worth a tenth of a normal users vote.
Categories of Articles
This could be expanded in such a way where every article has an option to be dugg. Users tend to be interested in a few areas of news and what they digg will generally show this. An algorithm could figure out what kind of section an article could fall under acording to that users history of dugg articles. An simplified example of this would be two users, 'A' and 'B'. 'A' digg's an article, and based on his history he has dugg 80% of the same articles that user 'B' has dugg, so based on this 'B' may be interested in the same article 'A' was looking at. Such an algorim when scalled to include hundreds or thousands of user can become very acurate and interesting for users to get there news. Each user's page would look different because there history of dugg articles would be different. The articles could be scoured from all over the web to cover many topics. Once a computer finds an article it starts off with zero diggs and an uncategorized section, but as users digg articles it could be categorized in it's apprate section. It may take a while for a article to be dugg to the main page, so for this case, user could change the web page around to show all articles by time written, or Article they would like based on the time written, or Dugg articles that are popular and are most recent.
A website that enables users to submit essays or other works to be peer reviewed. The more someone reviews others, the more likely theres can be reviewed. Feedback can be given to reviewers to have a good rating. The rating system would be similar to that of e-bays.
By having an algorithm go through Wikipedia and make connections by what is linked to other sites within wikipedia. Therefore similar types of products and people can be made to similar things. For example if someone searches Bmw they would see one of the "connections" as Acura, because this is on the page and has been liked by using "[[ ]]" brakets. The amount of time the same word is liked would increases it's popularity. And then also on the Acura page Bmw would be liking back to increase the importance of the link. This idea can be expanded into similar music types, actors, movies and find what a seprate user might be interested in from a users contribution list. The more words on a page would increase it's ranking. The amount of words on a page linking to another page would help it's popularity. If a lots of users edit article A, and more then half of them edit article B and C, then Article B and C are directly related, and Articles B and C, have a indirect connection with article A witch may be interesting to study.
This partly seams to exist: http://tools.wikimedia.de/sixdeg/index.jsp?from=bmw&to=m3
Personalized news. When a user clicks on a particular article to read, then it would look at other people who clicked on that article, then show you what other people looked at that you might like.
A website that displays the birds eye view of stores, so consumers can check the web site layout and go to exacly where they need to in the store. For example, it will locate the store you want, and show you in your local grocery store where they have the beer, bread and other main categories of food.
By using the Bittorrent software, a torrent web site can be made almost anything.
Legal torrents or free independent artists recordings and independent videos.
A torrent website that shows torrents with the most seeds all across the web. It would tap in to each tracker and grab the top 500 torrents with the most seeds. After getting all the from as many torrent websites as possible, then it can create a huge list of the most popular torrents. The torrents could be sorted out into each category for audio, video, etc. Users can see what what is the most popular, and then can see if it is any good. This could also be extended to the most torrents that have been finished.
A person or group who wishes to have a website made produce a design document and submit it to the site. Prospective developers read the design documents and post bids on websites they wish to build. Developer's bids state their proposed time for delivering the website, their fee for delivering the site, and other standard clauses commonly found in web contracts (IP rights, maintenance etc etc). The site is setup so that all bids conform to prescribed standards so that they may be directly comparable. Similarly all design documents must conform to prescribed standards.
The benefits of the website would be to allow medium and small organisations around the world (especially in europe or oceania where local skills are lacking) to obtain the latest web technology at competitive prices. Web developers who do not have cultivated a customer base, or do not have the oportunity cultivate one (perhaps the developer is in a third world country), will have easy access to a source of customers. Because design documents are transparent and standardized, the possibility for 'feature creep' is minimised; there is a better chance for the customer to get-what-they-want through use of the site.
The website promotes the separation of website blueprinting (database design, featureset etc), to website implementation (coding, etc). Another benefit is for coders to get quick work without having to concern themselves with non-coding responsibilities such as project planning, or customer-contractor relationsip management.
Customers would pay a premium to list their design document. Developers would pay a premium in the event that they are signed to produce a site.
Wiki Web site ideas can mainly be seen hear: Wiki.