Feedback and Movement

May 26, 2006

It's been almost a month since the first release and overall usage is unspectacular, but not disappointingly so. New things always take time for people to contemplate. However, it would be nice to get some form of external feedback and dialog. There's been a growing consensus within myself that I have a quirky way of doing things. I like doing things different, but at the same time, I'd rather not stray too far into never-never land.

In the absence of feedback, I've found that there are times when I'm not searching for anything in particular, but just want 'something'. Um, yes, that probably sums up a lot of Internet usage, but with respect to Fact200, I notice I've started using it as an RSS newsreader. Fact200 works really well for this because 80% of the stuff through RSS feeds is simply stuff I'm not really interested in. Instead of scanning through titles or summaries to determine interest, I can now scan through actual articles just as fast. In the cases of lots of REALLY random stuff (Craigslist), I can do a quick search to see if any new stuff matches points of interest. As an example, I ride the N-Judah everyday for work. Thus, a search for 'judah' in Craigslist Missed Connections lets me easily see any potential MCs (none so far 😦 ) and adds a Where's Waldo type of spice to the commute.

I've also run into the problem of not having Wi-Fi access on the laptop due to lack of hotspots (The Google/Earthlink deal doesn't make me too hopeful considering I despise Ads). A nice aspect of Fact200 is that it keeps all information locally stored and thus it doesn't need Internet access for either searching or viewing data. I like the notion of 'hot-syncing' with the Internet and then being confident that I have at least a minimal set of information that's always accessible. Webaroo offers this, however their webpack size makes 'hot-syncing' way too inconvenient, not to mention the data in their packs only provides spotty coverage of things that I'm interested in.

I'm thinking of using Fact200's category mechanism to define areas of interest where users can download rss feeds, search result sets, and static URLs all in one action and all from original sources (i.e., not prepackaged webpacks). There might might be one category set for doing research where the links/data are mostly constant. There might be another category set for news or local events that would result in constantly changing data. Because these sets are located on each person's machine, they can customize them to fit their needs.

Another aspect of the Internet I've started using more often is link sharing sites. Actually, I don't really share links (or search for them in these sites) with anyone else except for myself. It's convenient having a centralized site for links and I'm a little surprised that browsers haven't built in access to these sites. However one of the reasons my link list has remained static for a long time is that the whole tagging process is cumbersome. Most of the links in my list do not have tags associated with them.

The current method of accessing links in Fact200 is by searching for content. For sifting through a huge collection of links, this method is probably OK (and if you consider the Internet as a huge collection of links, it is fairly necessary), but when searching for a particular link it can be a little indirect. The library category helps alleviate this a little bit, but it's slow and suboptimal relative to bookmarks. Fact200 needs to work better with fixed, centralized link sets.

Advertisements

I thought patents weren't supposed to be issued for "obvious" inventions?

Here's a blog entry at SEW that notes a couple of details in some of Ask's patents that seem to touch  Fact200's previews a little.
Maybe I should brush up a little on my patent law…

A Little Odd

May 16, 2006

I didn't quite meet Scott Adams during a brief internship at Pacbell during the summer of '93, though I got a great tour of his cube.

I hope this never happens to me. 

Snap

May 15, 2006

Is it just me or has there been a sudden confluence of releases by search engine companies that happen to overlap Fact200? Alltheweb released last week, now snap.com (unrelated to snapfiles) released today. The major features of snap's release are term suggestions (like Google suggests) and full size previews.

Their term suggestions are way inferior to alltheweb's. It seems similar to previous incarnations of suggestions where there are a limited, pre-processed set of related terms.

The previews are nice. I think Snap had them before, but they've streamlined them a bit to make it as easy as possible to 'flip' between search results. They've created a setup similar to Fact200's keyboard interface where you can use the up and down arrow keys to change your selection and have the preview change as the selection changes. Even displaying the smallest preview (the largest being a full-size albeit extremely truncated preview) has noticeable lag due to network latency. To do the same in Fact200, you use the left arrow key to bring up the preview window then use up-down keys to change your selection and have the preview update. Oh, and another 'feature' that they're getting some attention for is their mixing of sponsored results into organic results. Read the story on here.

I've used snap before and definitely find the previews helpful and the biggest lacking feature in alltheweb's search results. It's nice to see all these server-side web sites re-affirm the qualities of Fact200 even though the competition is somewhat disconcerting.

Google Co-Op

May 12, 2006

Google seems to hard at work improving and innovating in their core proficiency: search. Google Co-op looks like two separate things. The first is subscribed links which is appears to be very similar to Fact200's sources and in general allows Google to become an aggregator of search results beyond Google's own index. The second is 'Topics' which appear like their attempt at a del.icio.us-like community based (err, sorry, "Web 2.0" for the trendy folk) form of editorializing.

Subscribed Links looks like a very simple method that any web site operator can get their results shown in Google's search. There is a catch though: People have to subscribe to your links. There's also some coding involved to adhere to the subscribed links xml format but this should be trivial and there's probably going to be open source support (in various blog and forums software) for it soon. What's a little odd to me is that subscribed links appear in the same spot premium sponsored links would show up. They essentially look like ads. I think this might be an attempt to re-sensitize users to their ads and boost ad value. But it also seems like this limits users to only one or very few subscribed links because it'll probably get jumbled really fast and also push the organic results way down the page. Actually, that's another factor in their implementation: response speed.

Google's own index is lightening fast. However, most other content providers don't have quite the clustering power of Google's search engine and thus the subscribed links will either slow down the response of your query (making Google seems slower than it should be) or timeout in which case the subscribed link won't show. Actually, I haven't seen it in action and it might be the case that they use AJAX to fill in the subscribed link section as soon as the external feed responds, but this might cause an unexpected and awkward (depending on how big the subscribed links are) shift in your search results.

Topics is interesting but rather…dare I say, confusing? At first, I thought it might be a lot like del.icio.us's tagging which is pretty simple. But after digging a little deeper, it appears somewhat similar to Fact200's categories and sources.

The way I think it works: 

You 'label' or 'annotate' URLs (Either I misunderstand this or Google needs to consolidate their nomenclature). This is pretty similar to 'tagging' at the basic level, but they support some advanced tagging features like attaching a relevancy score and wildcard url expressions.

Next, you create a 'context' file which contains facets, labels, and triggers. 'Triggers' are how Google knows to apply the particular context/topic. Facets are the subcategories Google presents when it detects you're in a particular topic and labels are connected to the facets which help narrow your search.

 The end result is a system very similar to Fact200's where there are logical groupings (Topics/contexts in Co-op, Categories in Fact200) that offer ways of narrowing down search results. Creating them is meant for the die hard nerds and geeks who can decipher the esoteric system (I'll admit this currently applies to Fact200 as well). They in turn, share them with their not-as-sophisticated friends who say "Golly gee, you are so smrt!"

Trying to remember back to an example of yahoo's attempt at real-time search, dubbed "instant" search, I finally found it here. It didn't look like it had changed at all since I first viewed it and their initial attempt was pretty paltry with an extremely limited set of search results. Scanning the forums, there were various mentions and comparisons to Google Suggests which also looks like it hasn't changed too much. But then, almost coincidentally, I stumbled across a post where it looks like alltheweb has just released a version of 'live' search here. Unlike Yahoo (actually alltheweb is yahoo now) and Google's experiements, alltheweb operates much more like Fact200 in both search results and term completion. Actually, they do a little more sophisticated term relationship management yielding a arguable better term suggestion mechanism. It is indeed, very impressive.

Doing a search around for 'live' and 'instant' search, it appears that a few others have been experimenting with real time search on their own sites (bitflux and a blog) and the major question that still remains is how well is it going to scale. The number of requests is probably at least 5x more. Hardware and bandwidth are getting cheaper though…

On the plus side for Fact200, they are still less responsive, lack previews, and lack indexing of ideograms. Fortunately, someone on the message boards made a note of their lack of Chinese support and provided a segue for a Fact200 mention.

But they do provide real time searching over a global search domain which is definitely interesting and worth keeping an eye one. 

The First Wave

May 10, 2006

Welp, it's been a week since the first release and there seem to have been roughly 500 downloads, with about 300 actual trials (a nice little perk of an update mechanism is the ability to get a feel for actual application usage). Not too shabby, but the forum is still empty and the main objective of this release, getting some feedback, is elusive.

So, its funny that I blogged on Webaroo earlier because I'm thinking that perhaps the 'offline' portable laptop audience might be a better target for finding users. The lukewarm feedback is an indicator that people really are content with current search mechanisms.  So, another potential area is the offline market (at least the Webaroo folks seem to think so, although the Copernic folks would seem to support the original idea) where the growing usage of laptops on-the-go might see a great benefit for having some portable content.

But, as a teapot once said: "We'll wait and see a few days more…"

Snapfiles

May 5, 2006

Welp, kudos to Snapfiles for only taking a few hours to post Fact200 to there site. They seem to get a little more traffic than version tracker as Fact200 already has almost 3x the number of downloads (90) in 1/3 of the time. The entry is here.
Snapfiles has a little more of an editorial process, ranking submitted products and also inserting their own screenshots and descriptions. This is actually pretty interesting as far as feedback goes. First, they gave Fact200 only 3/5 stars (boo!), but after some thought and after reading their description, it starts to make some sense. Their main criticisms were the UI and documentation. Documentation is pretty well merited considering there is none. Criticism of the UI is a little confusing, but from the screenshot, it looks like the reviewer was playing around with the Category/Sources functionality which indeed needs some refining.

But more disturbing is the seeming lack of interest and that it only states that Fact200 'has a lot of interesting features'. Hmph. I think what Fact200 suffers most from is a lack of accessibility. First impressions make it look like nothing more than an aggregator, of which there are about 3 bajillion of. I think I am going to make a Quick Reference sheet, both as the first form of documentation and as a little pictorial demonstration of Fact200's features. One of the problems of looking too similar to current tools could be that there is a lack of expectation.

Another problem is that many of Fact200's i capabilities manifest themselves over time. It's not very useful to provide real time search over a test scenario with known results. Nor is a library useful till it starts to get filled. Term completion also depends on the library having something in it. Lack of advertising annoyance is also a subtle side-effect that isn't really noticed.

The only thing that can really shine in the short term is the preview feature. Aggregation is also apparent in the short term, but it seems like most people are content with Google as a single search source. Still, there may be potential to sparkle a little with some of the unique data sources out there.

Cogito ergo sum

May 4, 2006

To highlight some of Fact200's features, I've been building a Philosophy library, during which I've stumbled across another 20(!) or so bugs and odd behaviours. Ohwell. How else should I keep busy?

But after scanning for a little bit there is a question on whether or not there should be some sort of way to 'browse' through the search database. I generally eschew the notion of browsing the database because I think that browsing should be left to a real browser like IE or Firefox, but it might be useful to somehow provide easier access to various search terms and allow users to explore the different terms without having to type so much. There are already 'hot zones' in place for search term completion, it might be nice to be able to click on terms in other places.

Or maybe, similar to some experimental web sites, provide a tree of search results to navigate down. This probably wouldn't work too well in the current UI.

In other news, I also discovered Fact200 still seems to hang on exit occasionally. This is starting to get annoying….I don't know how much more punishment the keyboard can take.

VersionTracker

May 3, 2006

Kudos to the guys at versiontracker.com who had Fact200's entry approved in about an hour. On top of that, the site is nice and straigthforward with no advertising clutter.

You can see Fact200's entry here