Google starts charging

February 25, 2007

Recently I wrote a post about how economic logic must be pushing us towards renting applications rather than buying them. Google don’t charge the user for Gmail, calendar etc.. but it seems they have just released enhanced versions for businesses for which they do charge – see this BBC news item.

Interesting move. Of course there is a long tradition for companies like SAP and Oracle to suck money out of corporations for “services” in an unending stream. Google’s charges are pretty modest by comparison. They are aiming at small businesses, as the Beeb says. So probably this won’t change the logic of the consumer market – applications free at the point of use, once you have paid your monthly fee to the ISP. Maybe rather than ads being the main income stream, Google aims to extract tribute from the ISPs.

Another treat for geeks – Cringely’s latest post is about replacing the internet. Some are gripped and others reckon Cringely has finally jumped the shark.


The Virtual Observatory, big computers, and the rental economy

February 17, 2007

The problem is, software is too reliable. Pause for spluttering. Start digression, intending to explain later .. I just found my trusty ole 1987 MacPlus in a dusty cupboard. The 1987 Mac Plus It works. The entire hard disc has 20MB. My word processing package, WriteNow, made great documents and took up just 128K. The MacPlus did everything I wanted just fine. How did we get to PCs with a GB of RAM, and vast application bloatware ?

In my mind at least, this connects up with the logic of the Virtual Observatory as I have been pushing in talks for years. The number one problem is not the number of bits in astronomical databases; its the number of databases, and their heterogeneity. Slogan number one is therefore Standards, Standards, Standards. The number two problem is the I/O and last-mile bandwidth bottlenecks; these grow slower than Moore’s law. Slogan number two is Shift the Results not the Data. You can’t download the database and hack your own analysis programme. The data centre has to provide professional tools for manipulating the data at source. So we need to move a service architecture.

When we first started this game, many people said this couldn’t be right, as the whole modern trend was away from old fashioned big computers at special centres, and towards more power on people’s desktops; and that this was spirit of the internet too, and peer-to-peer file sharing and all that trendy stuff. But where are we now ? The world is dominated by a few gargantuan computer systems – they are run by Google, Amazon, eBay etc. In the spirit of the Web, this stuff just feels like its “inside your PC”, but the truth behind this illusion is enormous corporate enterprise servers.

So why do we keep buying bigger and faster PCs ? I think we might stop …Expansion in storage has made sense, as we wanted pictures, songs and movies. But this need could stop; people trust Flickr to store their pictures. Expansion in CPU speed has been driven by gaming; but dedicated hardware like the Playstation is much better and more reliable, so who cares ? What about the expansion in application and operating system size ? To some extent this has been driven by the desire for using a computer to be easier, faster, and with more bells and whistles. But most people feel that things are way past the point of good-enough-thank-you. But those poor old software companies – how can they make any money if we don’t keep buying new stuff ? They have to work harder and harder to convince us the new stuff is really better and we want it. Who is going to buy Vista ? Not me. Can’t see the point. This just has to crack ..

So what is the sustainable solution? Obvious. The customer needs to rent a service, not buy a product. They have to make us pay for every use. We are getting used to this idea already … You don’t own that music, only the right to play it. Most video use is not download; you leave it where it is and play it again. Now we can all run our calendars and spreadsheets on Google, and never need to buy an application. Pretty soon the whole idea of owning an application will seem quaint… Java Web Start seems pointed in this direction too – click here and get the latest version, thats another 50p thank you. As an individual customer you don’t pay per use on Google, as their business model relies on the ads. But you can only get to Google by paying your ISP a monthly fee. Obviously a service-based OS must be up next. Who needs any more than a very thin client ?

So for the VO, the service architecture is driven by need. For the commercial world, its driven by economics. But we are playing the same game.

What’s really taking off now of course is internet TV. This is how Google will finally take over the Internet, as explained by Bob Cringely. I will do my own take on that in a wee while. Meanwhile, astronomy also is moving towards movies – astronomy in the time domain. I shall definitely have a go at that soon.. No coincidence maybe that Google have joined the LSST partnership


Help Find Jim Gray

February 8, 2007

No news yet on Jim Gray, but his IT friends have organised the searching of thousands of satellite images. They have stopped asking for volunteers for that task, but suggest other ways to help. You can check out the helpfindjim website.


Vista : Astronomer fails to get rich

February 8, 2007

Some cynical readers may feel that the above blog-post title falls into the “dog bites man” category, rather than the much more gripping “man bites dog” type of headline which those hungry for novelty always seek. But behind these words is a human drama of burning passion, lost opportunity, and secret cannibalism. Well, except that last bit. The recent news splash over the release of Microsoft Vista made me ponder on why Bill is rich and Andy is not.

Back in year 2000, I was one of a bunch of UK astronomers burning with desire to build a new large telescope dedicated to optical and infra-red surveys, and to get the taxpayer to cough up for it. Well, it worked, and the “Joint Infrastructure Fund” duly coughed up on Joe Public’s behalf. We debated long into the night on what to call this beast, spewing up many dull or clumsy acronyms until Mark Casali came up with the wonderfully evocative name of VISTA. (The Visible and Infrared Survey Telescope for Astronomy … of course the visible wavelength camera got cancelled due to funding problems, but we couldn’t call it ISTA because that would sound like we were fond of Egyptian mythology but dyslexic.) So … did Jim Emerson, Will Sutherland, Mark, myself, or any of the rest of the merry crew rush out and snaffle up all the domain name variants along the lines of blah.vista.foo etc ? Nope. (With the feeble exception of http://www.vista.ac.uk) A few years later, did the most powerful corporation on the planet announce that its new operating system had the same evocative name ? Yup.

On to the year 2001. With a different (slightly overlapping) set of chums, I launched the AstroGrid project, the UK’s take on the whole Virtual Observatory game. Very early on we realised that we needed a distributed but transparent virtual storage system. In AstroGrid, we use this partly as a staging post for asynchronous services, workflows etc, but we also saw it as a shared user facility, building towards that Virtual Community thing, and our Project Scientist (Nic Walton) suggested we call it MySpace. (“Hi John, I re-ran that query with the parameters changed like you suggested – the results are in MySpace, take a look”). The name MySpace is kinda tacky, but it stuck, and several hundred astronomers in the UK are now using it. So.. did Nic Walton, Tony Linde, myself or any of the rest of the gang rush out and etc etc ? Nope. Did a vastly succesful new social networking business called MySpace get going shortly thereafter ? Yup.

Sigh. So if only we had etc etc the above mentioned corporate giants would have swamped us with dosh, desperate for those domain names. Ahh.. now hang on .. I have this itchy feeling there is a flaw in that logic somewhere. Maybe they would have come up with different but equally good names … Hmmm. Hummph. Outfoxed by Gates again.


Jim Gray : mystery continues

February 3, 2007

Those of us in the Virtual Observatory world have been puzzled and saddened over the last few days by the Jim Gray’s mysterious disappearance. Jim is a very well known computer scientist and database guru, and over the last few years has made a point of working with astronomers, especially with Alex Szalay and his colleagues at Johns Hopkins. Last Sunday he set off sailing from the California coast, intending to scatter his mother’s ashes; since a cellphone call later that day, he has not been heard from. Seas are calm; no one has any idea what has happened. The coastguard has called off the search but his friends are not giving up. There are many reports all over the Internet.. here is a report from the Seattle Times earlier today.


Exploring the Cosmos Part II

February 3, 2007

So.. some days back in an earlier post I waxed lyrical on how mapping the sky has been of profound cultural importance – setting our lives in context, and satisfying a deep desire to discover whats out there. But are surveys the right thing to do scientifically ? When we map the sky, do we not spend many nights of expensive telescope time collecting data “just in case its handy” ? Would it not be more cost effective to do much more targeted experiments ? Au contraire …(Pretentious ? Moi ?)

Surveys are cheap

We do the science in two steps. Surveys are a kind of summarised digest of the sky. Once we have this digest as a database we can do multiple experiments from the same database, without having to go back to the telescope. This could be anything from “I have an X-ray source – does it have an IR counterpart ?” to “I want to test my model by calculating the correlation function of a special subset of galaxies chosen to have parameters XYZ”. This multiple use is hugely productive and very cost effective. In modern times the prime example is the Sloan Digital Sky Survey (SDSS). The team themselves have written plenty of papers, but astronomers all over the world suck data out of the SDSS archive every day, and the net effect has been almost a thousand scientific papers using the survey.

Put another way, online survey maps and catalogues can be seen as a kind of virtual sky. It is no coincidence that people like me who like surveys are also pushing on the Virtual Observatory front… more of this another time.

You need big surveys

Often the science we do just needs big swathes of sky. There can be three typical reasons for this. First, the thing we are studying covers a large fraction of the sky – like the Milky Way for example, or the Local Supercluster. Sometimes it is because we are looking for needles in a haystack – very rare objects such as quasars at z=6 or the very nearest and coldest brown dwarfs – we expect a handful over the sky, but don’t know where they are .. Finally, it can be because we need a very large sample – millions of objects – to beat down the statistical errors. The classic example is measuring the power spectrum of galaxy clustering, where we want a few percent accuracy in every spatial frequency bin. Another good example is mapping out the dark matter by measuring the distortions in galaxy images caused by weak gravitational lensing. (As in the recent result from the COSMOS survey). To do this, we need to average over hundreds of galaxies in each spatial bin in our map.

Surveys are the engine of discovery

Since the 1950s, every time we survey the sky at a new wavelength, we have been taken by surprise – radio galaxies, X-ray binaries, gamma ray bursts, huge starbursts. (More examples in the earlier post).

Many people would take this aspect of surveys as the best value for money for the last fifty years. We have been through a Golden Age of discovery. But has it now ended ? SCUBA, the submillimetre camera, plugged the last wavelength gap. It was very successful, discovering debris disks round nearby stars, and so many high redshift starbursts, that this seems to be the way that half the stars were formed. SCUBA2 will be bigger and better .. but not rdaically different. So is that it ? Have all the windows have been opened…. ? I will take a look at this in part III.