PracticeWriter Changes Ahead
If you are a traditional user of the PracticeWriter tool, we would like to hear your ideas or comment. Please review the situation described below and
let us know
what you think!
The PracticeWriter site was developed in the spring of 2005 with help from the regulars from the
Photo and HTML board. It was created mainly to provide a medium for idea exchange, troubleshooting and proof-of-concept for small sellers who needed to develop attractive ads or listings using basic HTML. The service has been popular and has been used by well over a million unique users who created over 250,000 documents during these last three years. Unfortunately, the accumulation of stored documents has become untenable. Scheming spam operators have automated the process of posting bogus documents and, even with filtering, the growth rate is hundreds per day.
In creating PracticeWriter, I initially ignored the idea of requiring user IDs. My thinking was that having to register for a service made it much less attractive. (Personally, I shy away from many tools simply because I don't want to take the time or I'm concerned that my email address might be sold or mis-used.)
Eventually however, not having user IDs proved a huge mistake. With no clear ownership of documents, maintaining the growing database came down to a choice between removing perhaps valuable documents or keeping them all. I settled on a stop-gap measure of allowing user registration as an option and spending many hours trying to associate meaningful documents with their owners.
Eventually spammers discovered the site and began to create off-topic documents to promote their own interest. Shipscript (from the
eBay boards) volunteered to help, and between the two of us, we have muddled along to "moderate" some of the trash but we are loosing the battle. Even with constant cleaning, the PracticeWriter database holds well over 100,000 documents and over a million session records which can't be safely removed because of the initial design flaws.
The result of this glut is that the database holding all this content is unmanageable for a low-budget operator like myself. It has become far too large to back up successfully, and, even though the web server is one the best available, it struggles at times to keep up with the load. Only load-balancing and multiple servers offer real performance for this kind of task and these kinds of solutions are beyond my means.
The solution would be easy if it were not for the need to avoid breaking links to meaningful
data. We could just empty the database and start fresh. But! PracticeWriter has been a popular place to publish files of all kinds, and judging from feedback and email, some of these samples and examples have been very helpful to those new to HTML. Links still exist in help pages and discussion threads to documents and, in at least three cases that I'm aware of, documents stored in this database are the last tangible work on the web from authors who have passed away.
The obvious goal is to pare down the number of records without breaking links to useful content and get a hold on future documents so that this doesn't happen again. Here is my solution..
- Restrict the persistence of newly created documents at PracticeWriter. By shortening the life of new records to a few days, we can at least get in front of the growth of permanent documents.
- Migrate legitimate users to AuctionWriter.
AuctionWriter is a ground-up re-write of the original PracticeWriter concept. In constructing this version, I had the opportunity to learn from the PracticeWriter experience plus start with a functional system of document management to relieve the database and server demands. By associating persisting records with the actual creator, spamming and off-topic data can be more easily controlled.
- Eliminate unclaimed and low-use documents from the PracticeWriter database.
During the next few months, we will encourage users with important content stored at PracticeWriter to migrate the document to AuctionWriter where it can be properly associated with the original author. Then, at a drop-dead date in the future, (to be announced) we will arbitrarily remove all low-use documents from the PW database and close the site to new long-life documents.