Elistic (http://www.elistic.com) started with an idea to improve upon existing services that allow you to share, find, and discuss links (like Digg, Reddit, and even del.icio.us) by making them more relevant to the site's users.

To do this, I wanted to incorporate a ratings system that allows a little more variety than just "good" vs. "bad", but is still simple to understand. Then I could use those ratings to analyze what users like and don't like, and recommend links to them based on other users with similar tastes.

From Idea to Product

Once I had this germ of an idea, I began to think about what form it might take as a web application. I started by opening a text editor and writing some HTML and CSS. I wanted to see what it would take for this idea to materialize. What would I show to the user? How would they interact with it? I didn't write any code at this point, and there weren't any images or a color scheme -- I just made up some dummy links and worked with the layout to determine what kinds of elements I would have to implement.

So now I had a pretty good idea of what components (in the general sense, not CakePHP components) would make up this app. They were things like user registration/login/account management, posting links, a star-rating system for links, comments, link retrieval (ranking algorithms, the recommendations algorithm), etc. From there it was a relatively simple process to start working on development. I created a simple database structure that grew as development progressed, and started right in on user registration.

Once I started writing code, I alternated time between that and the "real" site design. The design evolved quite a bit over the past couple months, and eventually settled on what you see now. It was a very iterative process, and I did the design work in chunks, using it as something of a distraction when I got tired of writing code.

In the end, it took about two months of evenings and weekends, working at a casual pace, to create what you can see now at elistic.com. I learned a lot in the process and hopefully created something that will be useful to a lot of people.

Challenges

There were (and still are) a lot of interesting challenges over the course of development:

Scope:

I'm just one guy working on this in his spare time. And I actually wanted to release this site at some point. So I didn't implement everything that I initially wanted to. I had to identify what features were absolutely necessary and what could wait, and then only work on the necessary stuff.

Critical mass:

This sort of site works best when lots of links are getting added and lots of users are rating the existing links. It's kind of a catch-22 -- if there isn't enough activity, people will lose interest and leave, causing even less activity. A critical mass of users is required that will generate enough activity to keep new users interested and to keep the website valuable to surfers. This is more a marketing problem than anything else, and I'm working on solving it by getting the word out to early adopters (hint, hint ;-)) any way I can.

Learning Curve:

This was my first CakePHP project, so I was learning the framework while I was developing the site. Thankfully, CakePHP isn't all that tough to learn, but sometimes I would implement something only to find there was a much better and/or easier way to implement it using CakePHP. That sometimes meant throwing out a piece of code and starting over.

Staying Ahead of Abuse:

It's not really a problem now, but if Elistic ever generates enough traffic, people will start trying to figure out how to game the system. I started by allowing users to flag links as spam, and automatically removing a link (pending moderation) if it gets flagged enough time. I also added an email address verification to discourage (but not completely prevent, obviously) multiple accounts.

But most importantly, I decided to make the ranking algorithms public (http://www.elistic.com/pages/how_elistic_works) -- the exact opposite approach of Digg, for example -- with the goal of working with the community to identify and improve upon "gaming" vulnerabilites (not to mention just plain making it work better).

Design:

I'm a programmer, not a designer, and I don't maintain any delusions of grandeur in that respect. It was a challenge to come up with something that looks good. I think I managed it, largely by paying attention to what sorts of things work on other sites, but I don't think I'll ever come up with something that looks great. If Elistic becomes popular, I will surely have to enlist the services of a professional designer.

Still to Come

There are already a lot of things in the development or planning stage that will present my next challenges. Some of the more interesting ones include:

Caching:

I've been careful to design the database to support intelligent indexing and minimize the expense of running queries, but that won't be good enough if Elistic grows to support hundreds of thousands (or more!) pageviews a day. I have yet to explore the caching options provided by CakePHP, but have some experience working with other solutions, and will be educating myself (and maybe designing some data warehousing-type solutions) as time goes on.

Site Tools:

I created a bookmarklet to facilitate the simple posting of links, but would also like to provide additional tools, such as a Firefox extension, and plugins for major blog software packages, that make it easier to integrate with Elistic.

API:

I'm a big fan of sites that provide APIs so that their users can provide new and unanticipated ways to interact with the service. As Elistic incorporates more features, this would become even more useful, so a web-based API is in the works.


Hopefully Elistic is a good example of what one person can come up with in a short amount of time using an excellent rapid development framework like CakePHP. I look forward to seeing you there!