Whenever I see the word "personalized" my hear skips a beat and a bit of vomit travels up my esophagus to annoy my sense of taste. But...

The suggestions are quite accurate, especially for Songs & Movies. In fact some of the recommendations contained some of my all time favorites, so kudos.

What's really important for you to nail now is the follow-up action: I want to be linked to resources that allow me to explore the recommendations, perhaps the media in question, perhaps more information on the artist or art.

Being able to reward myself for using your service with the enjoyment of an experience that confirms the accuracy of your recommendations.. would certainly make me come back for more!

I'm aware that everyone (at least your users that read HN) would be able to find the recommended media through a number of means (BT, p2p, Spotify, iTunes, etc), you should take advantage of the experience the user has when they are served such excellent recommendations, preferably by referring them to a service (which has the stamp of approval of your average user) which allows them to purchase/peruse the recommended media (or at least preview them).

Misc. ideas:

- Make #logo text (although I realize you'll probably won't have to rely on SEO for users), and even better, make it an anchor linking to the initial page.

- Put your text input on the same line as your "Go" button, and perhaps think of better copy for "Go".

- Give connect_widget_4e10ba13e07753d42974425 a bigger width to accomodate non-English texts

- Make sure your #how p DOESN'T break at twitter\n username, because right now, it does for me.

- Try not to use <font>, if only to comply with standards.

I hope my feedback is useful for you, good luck!

Feedback:

When showing the list of recommendations, keep the form with the inputs on top, so that I can change my option right there and see the next set. Going back (honestly I had to search for that link) and choosing again is tedious.

Also, for some reason, the twitter user name was gone and I've to type it every time. But I guess that could be my browser, not sure. (Linux 11.04, Chrome)

Wow thanks a lot for the feedback guys. Yes, it uses hunch. I could be wrong but hunch didn't have a way an unregistered user could just enter the twitter username and get the recommendations. Also, it was my first attempt at using apis and it was meant as a learning experience more than anything.

Does this actually work, or are the songs and movies just drawn from a list of songs and movies that most people think are great? I have a co-worker who's a hard rock / metalhead and his recommendations are indistinguishable from mine.

Interesting!

Haven't been using anything like hunch till now, so asking here. How does it work? Is is just parsing my tweets and recommending stuffs based on keywords? Where does the recommendation data come from?

EDIT: One more thing. It's pretty quick compared to some other services which try to read the tweets and do something with that. Is it really fetching and processing all the tweets before recommending?

EDIT: Ok, the above question was kinda dumb. Since it uses hunch, so it does not need to get the tweets any way. But, would be interesting to know how hunch does that, because considering that twecommend sends the request to hunch and hunch responds after doing all the processing, it is still quick.

Very neat concept. I found the recommendations quite accurate for me.

I second others in saying that you should try to integrate Amazon and iTunes affiliate links in there, to try and make some money. You deserve it.

Some suggestions:

* The text is a bit weird / difficult to read on my machine. (a 14" Windows 7 laptop, with resolution 1366x768, and running Mozilla Firefox 5.0). See screenshot here: http://imgur.com/5w6Ai

* Also, it would be nice if you could make the "Go Back" link a little more prominent.

Interesting. I tried it and it gave lists of magazines I read and tv shows I watch, probably because I have the word "prog" which is short for "program" in my twitter username.

Feedback: I have the same opinion as vijaydev, you should keep the form input on the top, it would be more convenient. Also, it's a bit slow, took me sometimes more than 20 seconds to load the recommendations. And possible bug: When you press on GO multiple times the loading image appears several times on the page.

Nice job, for a first application.

I'm not a prolific enough Twitter user, so I didn't expect the recommendations to be right for my name (they weren't close).

I tried using the names of some friends who do tweet/follow/are followed a lot but the site doesn't seem to be working at this time.

It's a neat concept/idea. Would love to hear a little bit of the background behind it.

It works well, and it astonished me with some movies, apparently it chooses the recommendations according to the people we follow on twitter (cf. Hunch).

Also, you should make it so that when you hit "Back" it keeps the username in the search field, it's a bit annoying to type it again.

Doesn't hunch already do this for me?

What a couple said: make the songs go somewhere I can listen to that song you are suggesting. If possible, within the same page so I don't have to come back to the app and play the next you. Kudos for you!

It worked very accurately for me. One suggestion - could you have the recommendations linked to their Wikipedia articles? this would save me from having to type them in.

I love these Show HN's: This is fantastic, please put a flattr button up! And also include some links to Amazon! It was pretty accurate to my interests too!

This is awesome. Why? Because it strokes my ego monumentally with a great list of songs. Put this on Reddit. It will get reddited!

It did good on everything but the songs for me. It would be nice if the username was still filled in when you hit back.

Nice! I suggest improving the "GO" button.

hi, I want to create a webapp as well but I dont know much about them, could you explain what technology stack you used and how you did things like deployment ?

great stuff !

It's down :)

Aha! Take Me On by Aha! is an old favorite song and music video. Your recommendation engine got this, so this is a +1 in my book.

You should look into linking available songs into the iTunes and Amazon music stores. Make some affiliate money for your project.