Showing posts with label google. Show all posts
Showing posts with label google. Show all posts

Thursday, April 02, 2015

Are there some things Google isn't telling us?

Here's a little something that's amusing but barely explicable about Google.


First I have to mention a couple of things about data modelling.  When dealing with data, an attribute tells you something about an object, such as its colour or date ordered.  In a dimensional data model, objects from the real world are set up with dimensions, which are essentially more structured versions of attributes, with multiple representations of each, such as day, month, date.  So in fact the terms are generally used in slightly different contexts.

Does that make enough sense?  Or can Mr Google find a way of explaining it more clearly?  Let's see:



So far, so good: it's anticipated my question.  But if you're a quick typist, you'll be most of the way there already.  At this point, Google somehow gets rather more... mysterious...


Dimension vs Alligator??

You may be asking, why not attribute at this point?  Someone discussed this recently in the back of New Scientist recently, in the context of predictive text on phones.  Inter alia, they noted that if you keep on typing, the algorithm figures the initial predictions are not what you wanted, and tries something... different...

Sadly, if you really wanted to know the difference between dimensions and alligators, Google goes all huffy, and compares alligators with crocodiles instead.  What are you hiding, Mr Google?



Tuesday, January 11, 2011

Evolution and monotremes, spam and off-topic Google searches

This post will discuss monotremes in evolution, but first a diversion to the reason for this topic.

I got a comment submitted from "Izumu":

"Hi Stephen,
I am Izumi from TBS TV, a Japanese TV company. We are intersted in the platypus egg photo you posted on your blog of 19 Mar 2008. I couldn't find your email address and that's why I'm making comment trying to be in touch with you. Could you kindly write me back to xxxxx@nifty.com ? Please don't post my comment since it includes my email address. Thank you very much for your coopereation. Izumi "

I post a reproduction of this comment sans email address, although I was inclined to include it anyway.

I get quite a number of spam comments posted, which is why comments are moderated.  I'm not inclined to reply to this request directly, because:
a) It was off topic;
b) The email address wasn't from an official TBS domain - Nifty is just a Japanese ISP.

Usually I just mark spam as spam.  I don't usually get a comment that's so close to falling either way.


There's a few pictures of platypus eggs on the web.  As it happens, mine is now at the top of Google Images.  It was a bit of a tragedy in some ways, because the actual topic of the post was the evolution of milk, but it gets caught in the wrong net.  If you want to communicate about platypus eggs, talk to someone who's communicating about platypus eggs.

The reason they appear to us to be strange is just a quirk of evolution: they are the last representatives of the earliest types of mammal.  The only egg-laying mammals (protherians) left are the monotremes, two species of echidna (porcupine-like creatures) and one of platypus, all native to Australia/Papua New Guinea.  Yet the first mammals were egg-layers.  Marsupials (metatherians: live but under-developed birth) and then placentals (eutherians: live birth) were a much more recent development, as the technology of birth evolved over tens of millions of years.

The oddness of the platypus may initially be due to their appearance, including webbed feet and a duck-like bill.  The fact that  they're mammals that lay eggs draws people in more.  But they are distinctive for two more reasons: they have poisonous spurs on their ankles (which seem to be for breeding purposes!), and they hunt through muddy water by sensing electrical fields.

The platypus, in evolutionary terms, is not so odd.  Pretty much all these features have evolved separately in other animals.  That's evolution: the time spans involved are so vast that if mutation can produce a lasting feature once, it can do it again.

No, the true oddness of the platypus lies in its survival to a time where most of its features are seen as uncommon.  There's a warning there: the survival of features that do not catch on (radiate) more broadly - in numbers or variety - is more indicative of desparately clinging to a vanishing niche than of evolutionary success.

Thursday, June 28, 2007

Searches: Warren Oates 3, Google 2 (Dave Graney 5 million)

Walkin' alone, drinkin' alone,
Ridin' slow, parked outside your home.
Sleeping just like everybody else.
There with no grace of god you go,
through the United States of Warren Oates.


Ahh, the Soft 'n' Sexy Sound of Dave Graney... well, in this case, one of his rockers, and a good one that I haven't been able to get out of my mind recently.


You might think that Google can turn up anything if you give it the chance. Put in a few consective words that quote, well anything... and if it's there, it will turn up. I do this a fair bit as a shortcut to details of what I'm after.

Now I'm not so sure.

Take the Dave Graney song above, The United States Of Warren Oates. Do a search on "200 miles of country road": 2 results - the correct answer, to my knowledge.

"Walking alone, drinking alone": none.

So the way Google indexes is not even. I suspect it breaks things up into phrases - for the less popular sites only - and only stores the phrases, not the full set of words. However, for the more popular sites, it indexes the lot.

Just something to watch out for. Try to quote a whole phrase, not two different but consecutive bits.

Over to you, Dave...


...A lead singer in a rock 'n' roll band,
you're poor,
you're stoned,
you're a slave to an unnameable half-forgotten ambition,
you're just another guy on the lost highway
a ramblin' man,
a pirate of love,
rider on the range,
a seventh son of a seventh son,
a love rustler,
a desperado,
a bastard right royal historically entitled to be bad,
the man in black
john da conkeroo
the world's forgotten boy
meat man,
mr blues,
the velvet fog,
the silver fox,
the little cloud that cried,
the best dressed chicken in town...


(Rock'n'Roll Is Where I Hide)


(...and it's where I hide too.)

Thursday, February 16, 2006

World: How to promote technological advance

The US Congress wants to keep Google out of China. To "promote human rights" - or to encourage technological innovation in enemy territory?


IT Toolbox notes that the US Congress is considering legislation to "keep vital computer servers out of China and other nations the State Department deems repressive to human rights". From what I've seen of Congress in action, this is a fairly typical small-picture approach. Next, watch them tack it on as a small item on a much larger, unrelated "must pass" bill. That's a one of many less-than-ethical tactics Congress has learnt over the years, which I have yet to see in other parts of the world.

Question 1: will this legislation foster human rights in China?
Question 2: will this legislation foster even Congress' version of human rights?
Question 3: is China at a pivotal point in innovation where this action would encourage R&D, resulting in a homegrown version of Google? (If so, the US would be fighting four-to-one odds, in terms of population.)
Question 4: just what does Congress really believe in, and do they express it with any consistency?
Question 5: Was a BBC commentator right when he called it simply a response to an economic threat, i.e. disguised protectionism?


I could keep coming up with questions on this. But I'll stick to one final thought:
If you think this is major cultural/ideological hegemony on the part of the US, I have one word for you: Hollywood.

You don't have to legislate hegemony.