Failure to Launch: Google Research Datasets
Google Research Datasets, we didn't even know thee. Consistent with CEO Eric Schmidt's pledge to cut experimental projects in the weakening economy, Google today, Dec. 19, confirmed that it is shelving Google Research Datasets, an experimental project for storing terabytes of open-source scientific data sets for life sciences, pharmaceuticals and other fields. A Google spokesperson quoted some of an e-mail sent to beta testers of the project:
Given what Schmidt has said publicly about axing certain projects, that's code for, "Eric decided it was a waste of time and valuable computing resources given the lousy economic climate and shelved it." Wired Science's Alexis Madrigal scored the scoop on the cancellation here, thanks to an e-mail Google sent beta testers of the project. The project was pretty hush-hush within Google, where it was code-named Palimpsests. Madrigal noted that Research Datasets was first revealed by Google engineer Jon Trowbridge at SciFoo (Science Foo) 2007. Here are Trowbridge's slides and here is a video presentation on the Archimedes Palimpsest data set. Some 30 data sets were uploaded to the Google Research site, including 120TB of Hubble Space Telescope data. That's potentially a lot of data that's no longer going to clog up Google's servers, which are taxed enough as it is. No wonder it was canned. Research Datasets is the third Google project to get 86ed in the latter months of 2008, following the closure of its SearchMash search results test and the abandonment of its Lively virtual reality program. Google's data centers are clearly getting a tad lighter. What other projects do you envision Google cutting in 2009? |

Comments (1)
At least Google knows when to quit.
Like little divining rods - the tiny hairs on the back of my head sometimes stand up when I see software architectures that show signs of becoming hopelessly bloated - and fragile.
Maybe it’s just the old country-boy in me – who’s quick to recognize the frailty and folly of Rube Goldberg machines.
I first sensed this 15 years ago with Microsoft's product line - which once again is showing its brittle nature.
With this recent IE fiasco where Microsoft actually confessed it might be wise for the public to switch to an alternate browser until this latest security flaw is ironed out – The BloatWare Factory is once again getting caught up in its own underwear – using overly complex programming methodologies & architectures.
Even before I started getting nervous about the boys in Redmond – I’d spotted another architectural nightmare starting up - back in the early ‘90s.
Three tier client-server architecture was to be the final nail in the mainframe’s coffin. I’d been considering re-working my career from host based to server based development back then – but decided first to subscribe to that top rated client-server oracle “Database Advisor” magazine - before doing anything rash.
“Database Advisor” remember them?
After a few years of reading that mag – it occurred to me just how terribly complex and fragile that architecture was. Every article purporting to off-load host apps to client-server was written by Phd’s who just happened to run their own consulting companies.
Never were there stories of how Suzy Q in some insurance company ported her app to client-server all by herself. There were endless articles about how you should hire network tuning specialists, mirrored database specialists, object oriented programming tool-management suites and on an on.
So just for fun I went out and purchased a single-user version of Powerbuilder – and found quite to my distress that a catalogue of third-party client-server “management tools” fell out of the box – which was twice as thick as the Powerbuilder manual itself.
Not a good sign.
Finally I decided to cancel my subscription to “Database Advisor” and skip the hype - and wait for the next big thing.
And sure enough – their next month’s issue contained a message from the editor – I kid you not – saying:
“We at Database Advisor have concluded that three-tier client-server architecture is not simple or robust enough for enterprise-wide business applications. So next month we are changing our magazine’s title to “Internet Advisor” and will be covering only HTML application development”.
And the rest is history.
I’ve been waiting for a similar fate to befall Redmond’s BloatWare Factory – and this latest fiasco with IE vulnerability – coupled with Google’s slimmed down browser approach and upcoming web-based business applications – may just be what the doctor ordered.
No doubt Chrome can be attacked too – but unless Google comes out with a raft of products like Active X – I doubt we’ll see a situation where Google cannot cope.
They won’t be using a mind-blowing legacy of bloated, overlapping and hopelessly brittle products like Microsoft.
Perhaps you could do some articles that discuss real-world examples of vendors getting caught up in their own spider-webs of software complexities.
Like the Wall Street Journal’s front page expose years ago about their discovery of a shadow Windows development team – mandated to secretly follow the primary development team and clean up their coding mistakes.
The WSJ’s visit to Redmond discovered this arrangement and how they found burned-out programmers in row after row of cubicles showing signs of stress and capitulation at the overwhelming complexity of Windows back then.
That’s back when Windows’ 35 million lines of code finally surpassed that of IBM’s mainframe MVS.
Or another WSJ front page article where SAP was contracted to replace a mainframe system at a major drug distributor in the US. They couldn’t tune the system to process more than 50,000 prescriptions a day – while the requirement of 2 million per day was easily met on their old mainframe.
SAP drove that company out of business and got sued by company conservators for one Billion dollars.
Here’s to the BloatWare Factory.
Rob
Posted by Rob | December 20, 2008 6:43 AM