I agree with you. I didn't build App Store Instant as a replica of Google Instant. With the APIs available, there isn't a reliable way to predict apps.
I built it because iTunes is slow. Try searching iTunes and clicking on an App (to read the description). It takes me 5+ seconds. This takes ~1 second.
Are you logging your queries? If you have enough searches you could build your own predictions. You're also proxying results, so you could use those to build an index too.
What we really need is a better app store interface on the iPad. Apple's implementation is terrible[1]. AppShopper shows it would be possible to do. I'd even pay money for such an app. The app could include "spotlight apps" too, as another source of revenue.
[1] It is terrible because it always resets itself to the beginning everytime you launch it. Couple that with the fact that installing an app causes the app store to close, means you just can't search around very effectively on it. Simple example: navigate to the third page of top apps, find an app you like, install it, go back to the app store, you've been pushed back to the first page and have to redrill down to page three to keep perusing.
Agreed. Browsing is especially difficult when you click on an app to find out more details about it, click back when you're done, and then find that you've been kicked back to page one.
Unfortunately I think this is one of the places where you're likely to get rejected by Apple for 'duplicating built in functionality', even if you've improved upon it, so I doubt anyone is going to do it
Although I've got "instant overload" too this is the first of the instant apps with a built in, scalable business model - each of the links on this site is iTunes / linksynergy affiliate enabled - that's truly smart :)
FWIW, you may want to test the linksynergy links through a non-US proxy. I implemented linksynergy on our app's marketing site to hopefully glean an extra 5% on the sales, instead we found there was a bug where international users were being dumped into an infinite redirect loop.
The issue occurred a few months ago but after losing a large amount of sales overseas before we tracked down the problem, we decided we would rather have better control over our sales channel than the extra 5%.
It would take us more than 6 months of sales via linksynergy to make back what we lost in the short time the issue was occurring. It's not worth the risk for us.
Also, it would be nice to find search results for iPad apps and to be able to browse beyond the initial 9 apps that are displayed.
Assuming you're not simply being facetious, what would that look like?
Some sort of meta-rss feed that delivers a 'best-of' for weekend perusal? Perhaps it could be collated/curated and then printed in a handy offline version for weekend digestion?
1. Look for weak signals (i.e. some random post about some random subject area)
2. Monitor to see whether it gains traction up slowly but surely (i.e. not your typical HN burst algo but something that might be months or years to pick up)
3. Deliver to user before it peaks but after it has reached some momentum.
The JavaScript (and a bit of the CSS) looks awfully similar to iTunes Instant which was released a few days ago: http://labs.stephenou.com/itunes
Edit: according to a few people I've heard from, Stephen (creator of iTunes Instant) did help out / supply some of the code for App Store Instant. My mistake. :)
Not to mention Domize (http://www.domize.com). I remember using them years ago, I still love it. It's funny, since I can't stand Google Instant, but really enjoy Domize, interesting how that works.
How did you manage to scrap the Apple Store? Are you storing all the apps data locally? I guess it would be a huge overhead to send each of these queries to Apple...
It's just a preview of the first 9 results. Apple only returns back a limited number. I need to find a better way of displaying if there are more than the 9 results. Any ideas?
OK, too bad Apple limits it like that. It really kills its usefulness then. Books in the App Store (aside from the ePubs in the iBookstore) is a rat's nest that's hell to wade through on the desktop. An Instant like this would really help with that.
I find the results often unexpected as a result of matching a search term to several app fields. I search the word "instant" and receive AIM as the top result (presumably because AIM is an "instant" messenger application). Same thing occurs due to matching against developer names, causing what I'd consider to be false (or 'less true') matches to show up because the string matches a developer name, and yet matching app names are lower in the match results, even though the string is in the app name--and often the beginning of the app name.
It'd be worthwhile to consider tightening the parameters. I'd wager a typical user performing an instant search is not searching by developer name. I'd further wager an average user is searching by keyword and description less frequently on an 'instant' search than by app name. So perhaps you could try a bit of weighting, pushing X app name matches higher in the results, followed by X description matches, followed by X developer name matches. Perhaps tweak the UI to reflect this.
And, of course, I could be completely wrong and this is all just a matter of my expectations not being fulfilled after searching.
very snappy, well done! it found all but one of my published iPhone apps, and very quickly, like Google Instant. Of course, one huge advantage you have is less traffic. And you've got less to render. You may also be showing more of a cached, less real-time view of what's actually in the store (guess).
I'm not supporting the above statement, and the blithe logic it displays, but I can't go along with this, either. If that is as deep as the analysis of motivations goes on Hacker News, none of you should ever be allowed to vote.
I'd love to respond to the below (dead) post and the thoughtful dorks who downvoted me, so might as well reply to myself.
I don't care why he wrote this application. My point is that you can't just say "LOOK HE SAYS WHY RIGHT THERE, THAT'S WHY!" and have any sort of intellectual credibility, unless the intellectual credibility of a third grader is what you're aiming for.
Partial queries are not useful.
My 2cents