Greg Linden, the CEO and founder of Findory, was nice enough to find some time in his busy schedule to do an email interview with me. As described on the Findory website, “Findory is about personalizing information”. Findory helps you make sense of news and information by recommending content based on past reading habits. You can read a more detailed Findory profile at TechCrunch.
This interview occurred solely via email over the course of several days. If you are on the go or too busy to read the interview, you can subscribe to my post-by-post podcast via Talkr by clicking on the “RSS Talkr” button on the sidebar or by simply clicking here.
What gave you the idea to start Findory?
Findory is about personalizing information. We are all feeling overwhelmed by the flood of information in our daily lives, and the problem will only get worse in the next few years. We need a way to filter, prioritize, and find focus. Personalization has been successfully applied to e-commerce where it helps people find books, music, and other products they may not have discovered on their own. We want to apply personalization to information in general, to help people find and discover the information they need.
Our first product is a personalized news site that learns from the news you read, searches thousands of sources worldwide, and builds you a personalized front page and news and weblog articles. It’s free and easy-to-use. Just read articles! Findory learns from the articles you read and helps you find other interesting articles. We also have early products in personalized web search and personalized advertising.
Findory was launched in early 2004. Was there any particularly big break that put you on the map?
Findory has been growing at a fairly steady rate of about 25% per month since we launched in January 2004. There wasn’t any one event that caused a major spike in growth. Since we are very small and have almost no budget for marketing, we have been fortunate to have received favorable coverage from local papers such as the Seattle Times, Seattle PI, and Puget Sound Business Journal as well as many prominent bloggers including Inside Google, Search Engine Watch, and Searchblog. Many people seem quite interested in the idea of a personalized news site that adapts to your interests. It’s unusual, something you can only do online, quite different than a normal print newspaper. It does make for a pretty good story.
Why do you think personalized news and information is so important? Why has it been so well received?
It’s hard to find the news and information you need. Skimming dozens of news sites or feeds takes a lot of time. Searching for news only works if you already know what you want or what is out there.
People need a way to cut through the clutter and surface the information they need. Personalization complements search. Personalization surfaces things you didn’t know about. Personalization helps you discover articles and news sources you wouldn’t have found on your own.
Are their any drawbacks to personalized content? Does it in some ways limit people by not exposing them to new ideas? (see iPod era of personal media choices may be turning us into an iSolation nation for what I mean)
The idea of a personalized newspaper has been around for some time — the “Daily Me” was what it was called in the mid-1990’s — and there has been some discussion of whether a personalized newspaper might pigeonhole people by only telling them what they want to hear. Findory had the benefit of hearing this debate and was built with the pigeonholing problem in mind. Our site explicitly avoids pigeonholing by showing a variety of articles around your interests. For example, if you read a few articles about events in North Korea, you wouldn’t only be recommended other articles on North Korea. Instead, you would see a broad variety of articles that interest people who read about North Korea — articles on China, Iraq, broader international news — as well as top stories that are generally popular or important. Yet Findory still targets closely to your interests, for example, showing articles about open source and Microsoft if you read about Linux, not just general technology top stories.
Recently, you added a feed reader to Findory. What prompted this move?
Findory is designed to be an RSS reader for the mainstream. Many have noticed that RSS is being used by many in the geeky early adopter crowd, but not by the mainstream, not by the grandmothers of the world. For RSS to enter the mainstream, we believe people shouldn’t even know that they’re using RSS at all. RSS is just a data format and should be hidden from readers. We think people just want to read news, so Findory makes it easy for people to read news pulled from thousands of sources worldwide without ever realizing they’re using RSS.
But we had received some complaints from power users who wanted more control of Findory. In particular, they wanted to be able to specify sources (e.g. BBC, Scobleizer, Wired) that they could read every day on Findory. Our new feed reader was designed for these power users, especially power users who are feeling overwhelmed by the tens or hundreds of feeds in their current RSS reader. Using Findory, the important news from their favorite feeds bubbles to the top, and power users can avoid or at least minimize the laborious processes of skimming the content from their hundreds of feeds to try to find something interesting to read.
Just to make sure I understand, are you saying that the way Findory works is by pulling in feeds from around the web? If so, what about sources without feeds?
Findory mostly crawls RSS and Atom feeds. We also do a little screen scraping for sources without feeds.
Being the small startup that you are, what is the biggest challenge that Findory faces going forward?
We are small, teeny tiny, just two people, self-funded. We’re the only ones doing personalized news at the moment, but all the search giants have talked about personalized news and information. It is daunting to be swimming with such big fish, the challenge nearly overwhelming. But with great challenges come great opportunity. We are building something no one else has done before. We are still ahead of everyone else. We are small, but we are quick and nimble. If we rise to the challenge, we should be able to use our size to our advantage.
What are your thoughts about the beta release of Rollyo? Is what Rollyo doing an important part of the personalized news and information puzzle?
Rollyo just launched recently. My understanding is that it is a search that you can configure to limit it to specific sites, a useful tool. Although this functionality is already available from other search engines, Rollyo puts a nice UI on top of it that appears to make it easy to search broad groups of sites.
As far as personalization, there are many other configurable search and news sites — My Yahoo, Bloglines, Google Alerts, My Google, My AOL, Start.com — but all of them require effort for you to explicitly customize the site to your interests. Findory is unusual in that it learns from your behavior. No configuration, no effort, no work. Just read articles and Findory gets better and better.
If you do not know who Peter Cooper is yet, chances are you will soon. Peter is the creator of FeedDigest, a tool that very neatly creates a way to include customized “digests” of live feeds in the content of your site (see the “New Internet Press” section on the TECHNOSIGHT homepage for an example). Dissatisfied with the slowness and approach of other options to put his del.icio.us bookmarks on his blog, FeedDigest (then RSS Digest) was originally a pet project that he eventually made public. His service now boasts of 9500 verified users – with no formal advertising to this point.
Although FeedDigest has paid for itself until now, Peter recently received an angel investment (amount and investor were not disclosed) that will allow him to work full-time “for real” on his once pet project . His immediate plans are to focus on the usability and interface of the FeedDigest service, expand the already impressive feature set, and ensure that the technical infrastructure will support the growing user base.
The complete text of the interview follows below. It consisted of Peter answering some questions I sent him via email and an informal chat via Gizmo Project.
Alternatively, you can use Talkr to listen to the article
Peter, you are the creator of RSS Digest (recently renamed FeedDigest), correct? Tell us a little bit about yourself – what has kept you busy to this point in your life? What is your professional background?
Yes, I’m the creator of RSS Digest, which has now been entirely redeveloped into the FeedDigest service.
My Internet life started out by doing Web design around 1996. I got bored of that scene pretty quickly though, and wanted to develop applications, tools, and services to bring Web sites to life instead. I started a Yahoo! style search index in 1996, but had to drop it after I realized I needed some Perl skills. A shame – it might have been big!
I spent a few enjoyable years during the dot com boom as a writer and content manager for startups that were mostly bought out by their larger rivals, and I came to work with the then giant Internet.com. Managing large Web properties gave me a good view into a world which lacked RSS, but was sorely crying out for something to fill the gap.
Throughout, I kept programming, both for fun and for profit, and now I’m effectively a full time Web services developer and systems administrator with my major interests in Perl, Ruby, SQL, Rails, Linux, BSD and OS X. So my background has been varied, but I think I’ve finally found my niche! The dissemination and management of information has always fascinated me, which is why I enjoyed being a writer so much. FeedDigest lets me hook up the technology of processing and managing information with the art of actually publishing it.
According to your website, RSS Digest has been around for a year but RSS has only recently really gained popularity. What prompted your creation of RSS Digest? How did you see the need for it before in many ways, RSS was even on the radar screen.
In terms of being a pioneer, RSS Digest certainly wasn’t. I’ve learned that my prime skill is to see where people are doing things wrong, and to build what users want in a certain niche. Companies like FeedBurner were in niches we plan to occupy way before we were, but we want to do it right. First mover advantage can stand for a lot, but I’m more interested in a “right mover advantage”. I want to get it right.
RSS Digest has now been renamed FeedDigest. What prompted this name change? Can you highlight some of the new features of FeedDigest?
The industry has opted to use “feed” to define what we know, technically, as RSS and Atom. FeedBurner and Feedster are two examples of established companies in the industry. Using RSS in a company name doesn’t seem to have worked yet, and to use it would deny the simplicity of “feed” as well as the existence of competing formats like Atom and, potentially, OPML. In order to become more commercially acceptable, I felt FeedDigest was the way to go.
Can you tell me a little more about FeedDigest and the “radically different (better) architecture to that of the current RSS Digest”? What was the architecture of RSS Digest and why was this change needed for the FeedDigest launch?
RSS Digest initially ran on a single Perl CGI script. It was okay for perhaps 10,000 requests a day, but no more. It was laggy, but workable. As people signed up, I faced an immediate need to rewrite it from scratch. RSS Digest 2 was born, and it still ran in Perl, but ran as its own server process to make it fast and lean. Over several months I added caching features and constantly optimized the code. It eventually scaled up to 1,000,000 requests a day before it began to show signs of cracking, by which time FeedDigest was released.
FeedDigest’s architecture is radically different. Its feed parser is still written in Perl, but parses feeds in a whole new way using a proprietary XPath based parser containing tens of “rules” of how to parse feeds. The parser is so flexible that if a new format comes along, or if I were pressed to support another format, I could have FeedDigest understanding it within an hour. Every part of FeedDigest is loosely coupled. Each part could, if necessary, run on different servers. The server process which delivers the digest, the user interface, the feed parser, and the feed cache are all independent and use normal Internet protocols to talk to each other. This gives us the ability to scale to 10 million requests a day and beyond.
FeedDigest is listed under your “Ruby on Rails” work. Many people are either unfamiliar with or have never heard of Ruby on Rails – can you quickly summarize it and why you choose it to implement FeedDigest?
Ruby on Rails is a Web application programming framework by Danish coding superhero David Heinemeier Hansson. In a non-technical sense, it’s like “programming Lego” for building Web applications. You still have to be smart and know how to design, but once you do, it’s a case of pushing the bricks together. At every stage you can see what you’ve built accurately, and you’re never under tangles of code. It keeps things simple and lets you focus on development rather than working around annoying limitations.
It seems that FeedDigest is a one man show, is that true or are you working with others on this project?
Until this point, I’ve been the only one working on RSS Digest and FeedDigest. No one else has even seen the source code. This should be changing soon, however, as the company begins to invest serious amounts of money into bringing FeedDigest up to enterprise quality. We plan to reach Technorati, MoreOver, and FeedBurner levels on a percentage of the budget, and we’re already a long way along that path.
Here is the million dollar question, how are you making money off of FeedDigest? Everything appears free – it seems that you host your own servers, do not include ads in the digests, and are continually making improvements to this tool. It seems like it is a full-time gig but are you paying the bills?
FeedDigest easily pays for its own technology expenses. It’s not making a living wage, by any means, but it makes a profit. 8 years of Perl experience, a deep knowledge of the industry, and Ruby on Rails have enabled me to develop most of FeedDigest quickly, while appearing to be working on it “full time”. As luck would have it, however, I’m now working on FeedDigest full time.
With FeedDigest now on the scene, is there any action that users of RSS Digest need to take? I noticed that recently my RSS Digest scripts were not loading on some of my pages. Are the servers that RSS Digest run on, which I think are Big Bold servers, going away?
No. Indeed, I upgraded the server RSS Digest uses just this week, and it’s a lot faster than before. RSS Digest users will be encouraged to move soon, but all migration so far has been by people keeping an eye on the RSS Digest site. FeedDigest has a long way to go, development wise, and I want RSS Digest users to be blown away even more than they have been so far.
Approximately how many people are using RSS and FeedDigest right now? Have you recently seen an increase in its popularity?
RSS Digest peaked at approximately 9500 “verified” users, that is, users with unique, valid e-mail addresses. As RSS Digest didn’t require e-mail addresses for the first several months, there may be up to 15,000 users in all, but I approximate about 12,000.
FeedDigest, which has a proper user sign-up and login system, has just topped 2400 new users in its first month of release. With no real promotion work done yet, other than seeding it amongst a few of my favorite bloggers, I’m delighted. The coming months are only going to get better as we now have a proper advertising budget.
Is there anything similar to FeedDigest out there? What is the closest “competitor” to it?