Author: JP

Sounds Like 2.0? Or, What I Really Want From Search

I guess I was younger and more hot-headed not that long ago; I remember being quite irritated when I heard that Microsoft had bought MongoMusic, as reported here. Until that happened, I’d nurtured real belief that we were on the verge of a breakthrough related to search, albeit rooted in music in the first instance.

Some of you may be aware of Shazam, particularly if you live in the UK. It’s a great service, you use your mobile phone to call them, stick the device near where there’s some music playing, the service samples the sound for 10 seconds and then texts you with details of the track and artist and so on. I love it, because it solves a simple and common problem for me. Say I’m in a car, listening to the radio, and there’s a song playing that I’ve never heard before, but that I like. Now I tend to listen to radio stations with very little talking, so what used to happen was that I was out of the car long before anyone announced what the particular song was. With Shazam, I could find out. Most of the time.

What MongoMusic tried to do went one step beyond Shazam; Shazam could only deal with formal recorded sound, “official” tracks. MongoMusic had a Sounds Like facility that allowed you to sing or hum or whistle the tune, and it would try and match it. So Shazam looked for exact matches, while MongoMusic looked for patterns, the sort of distinction that’s been driving image recognition experts crazy for some years now (in terms of database size and seek time and accuracy).

That was then. Now, reading a post in Smart Mobs, I see green shoots of recovery.  Take a look at this: Sing To Your Computer to Find Music, as reported by Roland Piquepaille’s Technology Trends, quoting Dr Sandra UitdenBogerd.

I look forward to hearing more, and to keeping all of you posted. I tend to think that this is really where search needs to head: where I can say Sounds Like or Looks Like or Feels Like or Tastes Like or It’s Somehow A Bit Like, and get a set of options to narrow down to what I want. I guess it’s what IdentiKit experts are used to doing.

Once we have something like that, imagine how we could help design things for customers. Identikit design backed by multimedia “It’s Like …” search.

Unwanted records

Looks like India will beat its own record for the highest Test innings score without a hundred: they’re currently 513 for 7, four runs behind 2nd spot and 11 runs behind 1st spot; if you interested, I published the top 10 such scores in this post. And it made me think, they didn’t really want that record. For sure Karthik and Tendulkar and Dhoni didn’t want it.

That took me on a tangent. There are many “unwanted” records in Test cricket, like for example the batsman with the highest number of 0s to his name. You may be surprised by the answer. I think it used to be Atherton, but I haven’t been able to check, Statsguru is too slow right now.

Why don’t you write in with your views on what should be the most unwanted Test cricket record? Seems a lazy August thing to do.

Not believing in our own propaganda

I’m a web 2.0 enthusiast; more particularly, I’m an enterprise 2.0 enthusiast, I guess; I’m sure some of you are as well.

Not everyone thinks the way we do. Which is why seeing videos like this one will help keep us honest and sane. My thanks to Nick Kings for the heads-up.

Incidentally, you can see the shape of things to come at the bottom of the Glumbert video “screen”. It says Send To A Friend. Digg This. Stumble It. Share On Facebook. Options you’re going to get used to.

On humour and vulnerability

I’m fascinated by comics, comedians, humourists, funny people. However you want to describe them. There’s something I feel I can always learn from them about the human condition. Which is probably why I enjoyed Not Only But Always immensely, Rhys Ifans does an incredible job of being Peter Cook. Man On The Moon was another I enjoyed tremendously,  about the life of Andy Kaufman. I still laugh whenever I watch Fawlty Towers, I’m that kind of guy.

Comics make themselves very vulnerable, it’s something about them that appeals to me. The same sort of feeling I got when I saw David Mackenzie speak before a private showing of Hallam Foe to bloggers, an openness and vulnerability that is rare.

I guess I can vicariously share in a tiny bit of that vulnerability by pointing to something offbeat that I found funny. Please humour me and go take a look at these robots fighting. As usual, also in my VodPod.

More on Build versus Buy versus Opensource

There have been a number of interesting comments on my recent post on the build/buy debate. Leo de Sousa feels I have missed out “reuse” as the first port of call. And I guess he’s right, since I haven’t been explicit about the role of reuse in all this. So I will modify what I said earlier and try and make this more explicit:

  • For common problems use Opensource (thereby reusing community property)
  • For rare problems use Buy (thereby reusing the learning of someone who has already solved the problem for a few others)
  • For unique problems use Build (but do so in an environment of component architecture and reuse)

Or…..

I could just say

  • For common problems use Opensource
  • For rare problems use Buy
  • For unique problems use Build
  • And in all cases make sure you maximise reuse

Abhijit Nadgouda points out that “with opensource, it is quite possible that you end up doing all three”. Which is also something I agree with, I think we are heading towards a time when opensource is the heart of all software development, but with some local builds and some specialised builds in a hybrid model. The distinction between build and buy and opensource then becomes one of scarcity economics versus abundance economics. Everyone’s got bills to pay, we just have to get the funding models right. Whatever the answer, one thing’s for sure, the current model’s busted.

Louis Nauges takes the argument in a slightly tangential direction, suggesting that SaaS plus open APIs is a different option. I guess I considered it nothing more than a variant of Buy, but then I may have misinterpreted what Louis said. Yes there are SaaS examples that are free-as-in-gratis, but opensource to me remains free-as-in-freedom, so I can’t really find examples of “opensource” SaaS. I need to think about this a little more.

Jag takes a completely different line, one that I am still pondering. What kind of ontology and taxonomy is needed in order for us to classify the problems and their domains accurately? How do we have to educate IT professionals as a consequence of this? More of this later.

Once again, do keep the comments coming. I really do appreciate them, even where I don’t answer back individually. And I learn from them. Occasionally I land up with quite a backlog as a result, but that’s a nice problem to have.

And remember, these conversations are snowballs. None of us should be proprietary about how they evolve and where they get to. You start them off and soon they have a life of their own. Sometimes someone else starts them off, and all you do is add bits to it.

Yes of course the conversations get fragmented. But this is better than stultifying and straitjacketing the conversation in the first place. And if we use tags and folksonomies sensibly, we can always de-fragment them.