CA's $1mn Open-Source Bounty Results

Anil Kandangath writes "Last year, Computer Associates open sourced their Ingres DMBS and they also announced a $1mn bounty for open source conversion toolkits from other databases to Ingres. Well, the toolkits are up on SourceForge and the bounty has been won by three teams, two from India and one from New York. More details and links to the projects on the CA news page. This is one of the greatest bounties for open source software and will hopefully serve as a model for other companies taking this path of cheaper development and better code."

Comments

There's an idea

[The Bungi]

1. Sit around and wait for a Fortune 500 company to issue a $1M bounty
2. Try to code a solution and hope you actually win (was: ???)
3. Profit!!

I think I'm going to quit my day job now. This looks like a great business model, not to mention an excellent way to pay the mortgage.

One milli-nano dollar?

[Bingo Foo]

What kind of bounty is $1mn?

$1 x (10^-3) x (10^-9) = $1 x 10^-12.

No thanks.

Bad news for two of the guys...

[Loco3KGT]

Two of those contestants worked for Oracle... ...and released a tool to ease migration from Oracle to CA's database.

Boy I hope Oracle doesn't hear about this.

Team India !~

[Anonymous Coward]

2 of the top 3 teams are from india, and the third entry from NY is an Indian guy.
Changing trends.

Cheaper, definitely.

[shmlco]

...serve as a model for other companies taking this path of cheaper development and better code.

Cheaper, definitely. Whether or not a team scrambling to meet a bounty deadline results in better code is open to debate.

Universities?

[starseeker]

I've often wished universities would do stuff like this. They have large scale software needs, (usually) a significant budget, and a lot of complex and fairly unique product requirements. I would think funding open source tools would appeal to them both in an economic and academic way.

Anybody who watched a Peoplesoft deployment at a university (and there were many of them) had to be both amused and shocked. I know my school spent millions - first to y2k proof an old system, then when that didn't satisfy them to go ahead and "upgraded" to Peoplesoft anyway. The result, at least from the student and professor point of view, was a nightmare. Buggy, klunky, and unpolished by any definition. I kept wondering why five or six universities couldn't have pooled their resources behind the GNU enterprise people. GNU enterprise + postgresql/ingres/whatever + other open web technologies couldn't POSSIBLY have done worse, and for that amount of $$ probably would have done MUCH better.

Heck, our CS students probably could have done better than the interface we got stuck with. It's no wonder college costs keep going up if what I saw was typical of university spending decisions.

Re:Universities?

[stratjakt]

You know what would also be cool? If they took your tuition money to Atlantic City and plunked it all down on red 36.

You don't know what you're going to get when you issue a bounty like this. It's a gamble. A good contract has obligations spelled out for both parties.

For every bungled deployment, there are dozens even hundreds that go smoothly. People just don't hop online and bitch when things work right.

Migrating applications..

[wfberg]

Sounds great.. Until you think about migrating applications; all those nifty stored procedures, never mind c or java tie-ins. The winners still have a long list of unmapped functions that aren't converted.

So, to what extent are these apps actually ready for the lime light, and to what extent did CA just choose a date to give away some money to grab some "free" publicity?

Also, it reflects quite poorly on all the databases (Oracle, DB2, and Ingres itself) that you *need* tools like this. If they could only have figured out how to stick to standards (or *jointly* come up with new, open standards) none of this would be necessary..

Wow...Has Tech Labor Truly Become so Cheap?

[ultimabaka]

I remember one particular scam I heard about when I first started looking for a job (I'm looking in Finance, but saw many CS programming ads as well) - it went a little something like this:

(a) Place job offer in newspaper
(b) Interview a bunch of candidates
(c) "Test" them all by making them write code to solve your problems for you while not being on the payroll.
(d) "Hire" one person, enjoy working code.

I can only imagine how much invaluable code this company got from making this $1m offer. I can guarantee you it was probably worth a helluva lot more than $1m. But, of course, none of the other entrants received a penny. This is just a glorified example of what I described above.

If this is the current state of labor in the programming sector, I worry and feel truly bad for you poor folks out there looking.

Re:Wow...Has Tech Labor Truly Become so Cheap?

[Jherek Carnelian]

(d) "Hire" one person, enjoy working code.

Sounds like an urban legend.

Project management is already such a complex process to get right when the developers are all under one roof and able to talk to each other that it would be nearly impossible to get anything remotely like a working system from the process you described. The end result would be more like a mish-mash of routines, all written with subtle differences "standard" input/out data structures and different assumptions about requirements and behaviour.

I can only imagine how much invaluable code this company got from making this $1m offer. I can guarantee you it was probably worth a helluva lot more than $1m. But, of course, none of the other entrants received a penny. This is just a glorified example of what I described above.

I doubt that much of the code they received was particularly valuable on its own. Sure it is possible that the code might be incorporated into another project, but it is more than likely that re-inventing the wheel would be easier then re-using code that was a) not written with a plan of re-use and b) the original developers are not even around to ask about how the code works and what kind of ways it expects to interact with other code or systems. Its pretty much an all or nothing proposition - either submissions get used for what they were designed for or they are going to rot away at the bottom of some DVD-R spindle.

Re:$1mn?

[winkydink]

One milli-nut. The amount of testicular fortitude shown by Anonymous Cowards on Slashdot.

Nice payday!

[PenguinBoyDave]

I'm glad to see that CA followed-through on this. While I am not sure how many people will actually migrate to Ingres, the fact that they put up the money, had non-CA judges review the entries, and gave them the recognition they deserve, to me anyway, shows that CA is making a good faith effort to show the Open Source Community that they indeed want to change the direction that CA has gone in the past. I see this as a good thing.

Surprise surprise

[interlingua.ro]

Contest rules:

the contest is intended for presentation in the united states, canada (except quebec province), mexico, india, china, the united kingdom, australia and new zealand. do not proceed within this site if you are not a resident of one of these countries.

(the lameness filter is lame)
No wonder the winners are from India.

from tfa

[frieked]

Only $550,000 was actually awarded out of a total pool of $1mn (mn? wtf?):
The winning projects were: Shift2Ingres, submitted by Harsh Azad, Rohit Gaddi, Achal Rastogi, Geetanjali Bahuguna and Ashutosh Upadhyay of New Delhi, India, won the largest prize of $400,000; EzyMigrate, submitted by Danes John and Varghese Jacob of Kerala, India, was awarded a prize of $100,000; and DbConverter, submitted by Bipin Prasad of New York, was awarded a prize of $50,000.

Here's links to the winning projects:
http://www.sourceforge.net/projects/shift2ingres
http://www.sourceforge.net/projects/ezymigrate
http://www.sourceforge.net/projects/dbcvt

From the original submitter

[alphakappa]

I submitted the story, so I should take responsibility for the typos there.
1. $1mn... stupid me, that should have been $1M.
2. DMBS... aah..dyslexia? well, that should have been DBMS.

Also, the reason why I said that this model will produce cheaper (obviously) and better code is that since it will be open-sourced, even if the original code might have taken shortcuts to make the deadline, it is still out there for anyone to tinker with and fix (if needed). And it almost guarantees continuous development.

Re:Topcoders vs CA Winners

[adbudha kusu]

Good observation but a closer look explains it. 1) Topcoder payouts are generally micro-payouts. And most of the decent Indian programmers are employed, and hence not "motivated" enough. The CA payout apparently was enough motivation. 2) I haven't looked lately but if I am not mistaken most of the topcoder payouts go to eastern europe. I suspect the decent programmers there have more time on their hands. As these countries catch the outsourcing wave, suspect their numbers on tc will drop accordingly. Yeah, i minored in freudian socialogy.