Quicksilver Roars (new features, Lion support)
lovequicksilver.com
lovequicksilver.com
One feasible path of exploration would be to use CMU Sphinx[2] connected to local NLP and Unity (or some underlying infrastructure that KDE and Gnome can share). But even if Shuttleworth hired a couple of developers now and had them getting some traction in 2012, you'd be looking at a usable beta for Ubuntu 13.04. At that point We'd be looking at the iPhone 6+ and one or two iterations of Android on top of ICS.
Microsoft has the base technology lying around in terms of speech recognition and Bing/MS Research/Kinect team and could probably come out with an answer on the next iteration of Windows (desktop and mobile). So you will have a first class solution on all Apple products, and a solid implementation on Windows as well as Android and perhaps Chrome. Nevertheless this forces Canonical's hand and they more or less have to move on this or they will be left behind.
My suggestion would be to take the following strategy (if Shuttleworth is reading): hire two or three developers or so for integrating Sphinx as a core service, and hire two or three developers to leverage NLTK on top of python (already the Ubuntu system language) and connecting it to Synapse. That's just four to six developers and two apt-gets being default rather than optional. Further, to save money, I'd contact a number of universities in the UK (Canonical is in London, logistics and legal are thereby simplified) to sponsor a few Masters and PhDs on basic research in the area with the proviso that their research goes open source and is patent free. In this way, a developer salary for one year turns into three years of research from a PhD. As I understand it they already have some people working on Kinect-like projects, but it needs far more resources otherwise it's just left to Google, who more or less don't care about the Linux desktop and tie the Android stuff to the cloud. Otherwise Ubuntu with a few tweaks is going to get slaughtered (if there is even one percent of market share to actually slaughter) by a cohort of competitors looking like you'd imagine OSX"11"/iPhone 6 and Kinect 2/Xbox 4 to be. The only other option would be to rely on the Google API from a native app, which would mean you could probably deliver something in Ubuntu 12.10, but you'd be totally beholden to Google then: and that's only if they wanted to let you use their servers with no ads going the other way.
[1]http://www.omgubuntu.co.uk/2010/11/synapse-gnome-do-launcher... [2]http://cmusphinx.sourceforge.net/
Also, Quicksilver only indexes the things you tell it to, whereas Spotlight by default indexes the entire system. For certain searches this leads to a lot of chaff in the search results.
Spotlight (and pretty much all Apple software) is rarely explicit about advanced (?) functionality but quite often there is some depth.
My favorite example (also part of Spotlight) is the calculator. It seems simple but it is capable of quite complex calculations. Try using sin, pi, e or sqrt. (If math isn't central to all you do this calculator is the only one you ever need.) You can also copy results by pressing Command-C. None of this is obvious. I didn't know that you could copy results for years, despite always complaining about the lack of that feature whenever I talked about the Spotlight calculator. (I was finally corrected that you can indeed copy results here on HN. Thanks guy/girl!)
I don't know whether that's a good or bad thing. I personally like software that doesn't try to please advanced users at every turn, but that's more about aesthetics than usability and I wouldn't try to generalize it.
I think you can guess from this that I don't use Quicksilver. Spotlight serves me well.
however, i much prefer quicksilver; it does a lot more than spotlight.
It also led me to my other issue with Spotlight - it can and will update it's list of found items, even changing which item is selected.
I've experienced this race condition quite frequently, and it's very annoying to have the right thing selected, then Spotlight swaps it out from under you and you've just opened something else.
QS doesn't have this problem, as it only changes the search results after receiving user input.
I'm glad development has resumed for QuickSilver, but I'll wait and see before giving up the reliably updated LaunchBar.
I've been running the previous version of Quicksilver on Snow Leopard and Lion, and it's been very solid; the new version should be even better.
i switched to quicksilver from launchbar, in spite of the fact that i'd paid good money for launchbar. it's not so much that something was wrong with launchbar; it's only that quicksilver was noticeably better in many respects (flexibility, esthetics, flow), imho.