Riya: new plan and Mac uploader
Funny, I was under the impression that this was the plan all along (although Munjal previously made a point of not wanting to be 'image' or 'visual' search, but specifically photo search)...
So, it is big news that Riya has shifted their direction to be a visual search engine, but the smaller news, which is closer to my heart is that there is a Mac uploader now available!
Yep. Finally! Anyone who has been reading me for a while knows that I've been complaining about this for months. I was also the biggest pain in the rss at Riya on this subject.
So, how does it fare now that it is out? Well, although I wouldn't recommend chosing the iPhoto option (doesn't allow me to choose subfolders, which sux), it seems to run pretty nicely on my non-intel G4. A little slower than the Flickr iPhoto plug-in or 1001, but that's because it's doing a good deal more work (that whole face detection thing that Riya is kinda known for. ;)).
So...Mac enthusiasts...have at 'er! ;)
On the subject of the re-focus of Riya, some people are concerned that this shift to image search indicates that Riya is waffling. I disagree. Face recognition has always been a means to an end IMHO. Sure, it's pretty cool technology, but eventually, you won't see face boxes at all. Face rec / text rec /pattern rec /etc. will all run in the background and you will just see good results.
Remember when Google came along years ago and we had no idea why the results were better, they just were? And so we all started moving over to their search because we found what we were looking for? There is all sorts of heavy technology running in the background that, if made explicit, would distract from the fact that you just got results you wanted. Period.
The issue that Riya had all along...and we had many discussions about this internally...was that by focusing on the Face recognition, the ultimate goal of creating a kickass photo (now visual/image) search engine would distract everyone. And it did. This is a lesson in long term vision.
When you want to create a long term vision, you should always focus on that, rather than the short term glories or gimmicks. Our bad. We knew face recognition would get people's attention. It did, but at the cost of creating a different impression of what we were really about. I remember very early on sitting down and discussing the importance of coming across as a photosearch site, which has a ways to go (building inwards, out...). Riya still has time to change that perception, but may get some flack for it in the meantime.
The major shift is that, instead of building from the ground up with trusted data (people tagging their own photos, leveraging the face recognition to auto-tag), Riya is going for the mass crawl. This I don't necessarily agree with. I agree that certain trusted sites should be crawled (RSS feeds, like Pixsy is doing + deals with other photo-sharing/storage sites on crawling their content). However, a web-wide crawl is just going to create quantity, not quality.
And relevance, in the end, is what really matters to a search result - not that there are 25 pages of results. You want the right results on page 1...at the top. Theoretically, you don't want people to have to visit page 2.
So, I'll be interested to see how Riya handles relevance with this new volume. That will be what determines the success of this "new strategy".
So, it is big news that Riya has shifted their direction to be a visual search engine, but the smaller news, which is closer to my heart is that there is a Mac uploader now available!
Yep. Finally! Anyone who has been reading me for a while knows that I've been complaining about this for months. I was also the biggest pain in the rss at Riya on this subject.
So, how does it fare now that it is out? Well, although I wouldn't recommend chosing the iPhoto option (doesn't allow me to choose subfolders, which sux), it seems to run pretty nicely on my non-intel G4. A little slower than the Flickr iPhoto plug-in or 1001, but that's because it's doing a good deal more work (that whole face detection thing that Riya is kinda known for. ;)).
So...Mac enthusiasts...have at 'er! ;)
On the subject of the re-focus of Riya, some people are concerned that this shift to image search indicates that Riya is waffling. I disagree. Face recognition has always been a means to an end IMHO. Sure, it's pretty cool technology, but eventually, you won't see face boxes at all. Face rec / text rec /pattern rec /etc. will all run in the background and you will just see good results.
Remember when Google came along years ago and we had no idea why the results were better, they just were? And so we all started moving over to their search because we found what we were looking for? There is all sorts of heavy technology running in the background that, if made explicit, would distract from the fact that you just got results you wanted. Period.
The issue that Riya had all along...and we had many discussions about this internally...was that by focusing on the Face recognition, the ultimate goal of creating a kickass photo (now visual/image) search engine would distract everyone. And it did. This is a lesson in long term vision.
When you want to create a long term vision, you should always focus on that, rather than the short term glories or gimmicks. Our bad. We knew face recognition would get people's attention. It did, but at the cost of creating a different impression of what we were really about. I remember very early on sitting down and discussing the importance of coming across as a photosearch site, which has a ways to go (building inwards, out...). Riya still has time to change that perception, but may get some flack for it in the meantime.
The major shift is that, instead of building from the ground up with trusted data (people tagging their own photos, leveraging the face recognition to auto-tag), Riya is going for the mass crawl. This I don't necessarily agree with. I agree that certain trusted sites should be crawled (RSS feeds, like Pixsy is doing + deals with other photo-sharing/storage sites on crawling their content). However, a web-wide crawl is just going to create quantity, not quality.
And relevance, in the end, is what really matters to a search result - not that there are 25 pages of results. You want the right results on page 1...at the top. Theoretically, you don't want people to have to visit page 2.
So, I'll be interested to see how Riya handles relevance with this new volume. That will be what determines the success of this "new strategy".
9 Comments:
Tara to Riya = Marissa Mayer to Google!?:)
Marissa Mayer makes decisions at Google.
LOL...yep.
Now, more than ever, I doubt if my opinion makes much of a difference at all. ;)
Unfortunately, Nick is right. And it really is quite a pity.
And I would suggest that, actually, a good visual search will provide you results not counted by pages or relevancy but by goal. When I do a search for something visual, my goals are varied... and if Riya is going to be a "visual search" it needs to go more than Googlify image results. It needs to help me accomplish my goal, whatever it is.
With Google, the problem was simple: give me the information I'm looking for. Seldom is that information actionable or worthy of being considered a kind of "oracle". If you want to buy a camcorder, why can't you go to Google and ask, with its infinite knowledge of your search history and email, "Which camcorder should I buy?"
Since Google can't answer that question, I wonder how Riya intends to meet the various personal reasons that people will use a visual search. In terms of interfaces to consuming visual results, everything but pr0n remains an open opportunity to meet people's needs.
Very nice change to your blogscape. And now I can read it without squinting!!
I knew that you could do it!
Bob Gagnon
Vancouver who didn't like your pastels
Nice new blogscape. I knew you could replace pastel shades with a new readable blogscape.
NICE!
For me, one places an awful lot of trust in a search engine to bring back the results one wants. From the precise words we type, we expect the engine to bring back the site containing the precise words we typed in the context in which they were written.
Now, move that on to images where there are hundreds of different colours and permeatations of colours and pixels, all merged together to create some sort of content, normally only perceptable to the human eye...then find a way to find search it, bringing back results with the same relevance as a purely meta-search.
Hard.
The magic with Riya and visual search is that there is now a fairly reliable tool which can understand and replicate how the human eye pieces all the dots and colours together in an image and bring back results based on the human identification of content.
I guess the next trick is how to make the Riya visual results better than other in the same way that Google trounce Yahoo, Alta Vista etc. for text search.
Looks like no love lost between munjal and tara.
Tara should should not rant about Riya and Tara in a public forum. That's just low. Munji's vcs are not going to allow him to create a new headcount as the "tara counter blogger".
Heh...
Love those anonymous comments.
I also love riya and where it could go. I was always and will continue to be the voice of the community for them.
If someone who cares can't speak up, then they truly will be in trouble. I sincerely hope that isn't the case.
I very much respect Munjal.
Post a Comment
<< Home