About Hi1: An Introduction

So What's This Hi1 Thing?

Hi1 is a project to build a search engine where users can type in any question (or search query) and get an answer returned to them immediately.

How do I use it?

Use it like a normal search engine, but be sure to give the best result(s) on the search results page a happy :) rating.

How does it work?

Hi1 is the meta-search engine that anyone can contribute to! Developers can write programs called 'engines' that specialize in answering particular kinds of queries. Hi1 then combines the output of these engines, and rewards those engines that perform well. We’re bringing together thousands of people to build the ultimate publicly accessible brain.

Here's how it works from a developer’s perspective...

1) A developer writes an engine and hooks it up to Hi1's query servers.
2) The engine will receive Hi1 users' search queries.
3) For each query, the engine generates zero or more search results, attaches bids to each result, and sends the results back. Whenever a result makes it to the search engine results page (SERP), its bid is deducted from the corresponding engine’s balance. The SERP shows the top 10 results, with each position going to the highest bidder for that position.
4) Any result that gets a happy :) rating will win a fixed amount of money, called the AR (amount rewarded). Whenever a link inside a result is clicked, that result automatically gets a happy :) rating.

Thus, a result's bid is proportional to the probability that the user will like the result, having seen other results on the page. So, if you write an engine that outputs relevant results at proper bids, your account gains money, else your account loses money.

Unlike current search engines, which depend on one overly generalized algorithm to answer every conceivable form of query, Hi1 divides the work among thousands of independent engines. Engines compete against each other (by trying to outbid each other), thus driving innovation. Engines also specialize in their own niche (e.g., in a particular type of search query, or in a particular kind of website). Although each engine will only have local knowledge of its own specialty, tying thousands of engines together will produce a superior search engine.

By being a platform, Hi1 will allow developers to test out thousands of different search algorithms in parallel. Current search engines usually delegate the development of new search algorithms to a specific team. Thousands of independent developers across the internet will come up with many more innovative algorithms than a single team, and the best algorithms will "evolve" out of the system at a much more rapid pace.

Hi1 is also fairer. The protocol is open, as is the ranking algorithm. Everyone has a fair shot at getting their site onto the SERP, so long as it will satisfy the user more than others can. If you're worried about spammers and/or irrelevant results, see the section called "How does Hi1 prevent outside market forces from affecting search quality?" (next page)

Register a Hi1 account, which will contain a free $10 balance for you to experiment with.

Why Implement a Hi1 Engine?

1) It's fun!
2) You're making the world a better place by contributing to a great answer engine project...how cool is that??
3) If your engine works well, you may potentially earn a lot of money (we plan on eventually converting your account's balance to real, withdrawable money).
4) You retain control of your code - make it open source or make it proprietary - sell it, reuse it, or throw it away - everything is up to you. We cannot view, sell, edit, etc. the inner workings of your engine since everyone hosts their own engines (for now).

Examples of Engines

• Engines that return results with the answer right in the body (e.g., dictionary, thesaurus, weather, sports scores, calculator, whois, language translation).

• Engines that specialize in a particular type of query.

• Engines that crawl a small subset of the web. For example, you can crawl one particular website that you know very well (you know the website's trustworthiness and reputation, you know where everything is located, you know how often it changes and what parts change the most, etc.).

There are hundreds of open source crawlers available on the web. Developers can incorporate new search algorithms into this existing code and try it out on a small subset of the web, expanding the crawl to more websites if the algorithm proves successful (successful engines bring in money, more money allows you to buy more hardware, and more hardware allows you to handle more queries and a wider crawl).

• An engine that answers queries by collaborating with a website and reading its database (that is, an engine that can search the "deep web").

• Human-Powered Engines, which use manual input from human users across the web to generate search results.

More Advanced Stuff...