Mark Shuttleworth Offering Bounties to Mozilla Developers
Wednesday December 3rd, 2003
Mark Shuttleworth writes: "I'm offering some bounties to Mozilla developers that are looking for small projects to work on. All work to be given to the Mozilla Foundation. For details please see http://www.markshuttleworth.com/bounty.html and follow up with Thomas Black at The Shuttleworth Foundation. Most of them are for work on Thunderbird/Firebird, but if the fixes also apply to the Suite then so much the better."
#1 Mark Shuttleworth Offerring Bounties to Mozilla De
Wednesday December 3rd, 2003 2:32 AM
The bounties are (currently?) for three bugs: 1.a. File e-mails in subfolders via keyboard (no mouse needed) b. Autofill folderlist (similar to "To:" field in message compose) 2. Bayesian filter to suggest most likely target folders. 3. Show error pages instead of dialogs for network errors (bug 28586).
It seems 2 and 1.b are related. The presented folderlist could look like this:
[ work ] <-- entry field +---------------------+ | most likely match 1 | <-- bayesian matches | most likely match 2 | (even if no letters are entered?) | most likely match 3 | <-- limited list length (to three)? |---------------------| | direct match 1 | <-- e.g., work | direct match 2 | <-- e.g., work clients | direct match 3 | <-- e.g., work clients marketing | ... | +---------------------+
PS. I wonder if Mark would be willing to "fund" other needed bugs...
Thank you Mark for offering to make this very generous contribution to Mozilla. :-D
#2 Re: Mark Shuttleworth Offerring Bounties to Mozill
Wednesday December 3rd, 2003 2:33 AM
so much for ASCII art in mozillaZine. :(
For those of you who don't like clicking links, here are the full bounty challenges for Mozilla (as of 2003-12-03):
* Thunderbird Message Filing Quick Access It takes too darn long for me to file a message. I use quite deep nesting on my IMAP folders because I have hundreds of mail folders. To file a message in the correct folder I have to click on the icon to file a message (or right click on the message) then click on a whole series of nested folders, for example: <email@example.com>, Inbox, mail, people, m, megan elliot. That's seven clicks, all of which have to be accurate, to get the message filed. If a click goes astray, the message is filed in the wrong place. To make things worse, the mouse has to be moved carefully through that hierarchy of menus, or the whole thing collapses and one has to start all over again. It drives me nuts. I am looking for proposals for programmers who will (a) enable key-stroke based access to the existing functionality so that message filing can be done entirely without the mouse, and (b) extend the current system with a sort of "folder type-ahead find" so that I could hit a hot key and start typing the name of the folder I want, and have it produce a smart drop-down list of matching folders, allowing me quickly to select the correct one and file the message there, all without using a mouse. This should be done as a Thunderbird extension.
* Thunderbird Message Filing Intelligence As an optional extra to the Quick Filing described above, I would like to see Bayesian learning for message filing. For example, I generally have a folder for correspondence with a person or a company, or a project. When I press the filing hotkey described under Quick Access above, it would be great if Mozilla Thunderbird automatically pre-populated the dropdown listbox of folders with its best guess as to where I want that message filed. It could use the sender, the subject and the body, and learn from previous filing decisions. Bayesian filtering of junk mail has worked really well for me in Mozilla, so I would hope that after using it for a while, it would be making fairly good guesses as to where I want to file a given message.
* Browser Error Pages (Bugzilla Bug #28586): $500 bounty Error dialogs slow down the browsing process, especially when one is using tabbed browsing and opening up multiple links in background tabs to be read later. There has been some work on replacing the dialogs with error pages, but this work needs significant improvement before the default behaviour can be changed. Given that work has already taken place, I would like anyone who steps up to the plate on this bounty to consult with people already involved to figure out a fair way to divide the proceeds, please send details of that discussion to Thomas when you claim the bounty. It might be best to let Thomas know that you are working on it in advance.
All the bounties I know about seem too small to get Mozilla developers' attention, even though the economy has not been kind to tech workers in the past few years. Have any of these bounties led to a bug being fixed or a feature being added? If not, I think the bounty offerers should up the ante before the economy improves more.
I don't know, he has offered a pool of US $100,000 (actual amount to be paid for each bounty from that $100,000 is to be negotiated with the person who offers to write the code) which seems rather hefty.
The only actual dollar amount I see on that page for a bounty on a Mozilla feature is $500. That would be enough to pay a consultant for maybe a day. Whoever writes that code must either be deeply familiar with the code in question or near starvation for it to be worth the work! Maybe some other bounties will be more realistic.
It seems like I'm not the only one who thinks the bounty needs to be much bigger before it attracts interest: <http://bugzilla.mozilla.o…how_bug.cgi?id=28586#c269>
How about fixing the bayesian spam filters? Currently it marks un-recognized words with a 0.4 correspondance to being spam. This should be changed to 0.5 or maybe a little higher so that spammers can't get away by adding junk random letters to an email. asdlfj l aoij oixcwe siodj ijewpwe kjwo ewkl oucdojw oie rlkqj er ofu qi qeijer oiq wekjq ddiao qwkj odou.
This is fantastic! We all owe Mark Shuttleworth our gratitude.
Let's not make this a one-time thing, though. The offering of bounties and microbounties should become a standard feature of Mozilla.org.
This is the next big thing. Let's make this happen.
The Mozilla folks seem to be happy for third parties to do it, but if Mozilla itself did it, it could cause problems. If you take a large group of people and specify some task in a not-very-detailed way and then say you'll pay money to whoever completes the task, you're going to get arguments.
The Mozilla folks decide what code goes into Mozilla and what code doesn't. If they are also deciding who gets paid for which bits of code, that makes it easy for conflicts of interest to arise, and arguments to happen, and Mozilla getting involved in arguments between volunteers about money could cause serious problems.
To make this kind of thing work, you would have to be very clear in specifying what was required and what kind of code was wanted, and have procedures for making decisions, appealing decisions, resolving disputes, etc etc. You'd probably need lawyers involved.
The folder filing stuff Mark has asked for as an extension, which should be ok as it can be self contained. The other stuff he's asked that anyone that's going to work on it should figure out how to divide the bounty in advance. Given that there are already several people involved, that's not going to be a trivial task, and someone at the Shuttleworth Foundation is going to be looking at it. Administrating most micro-bounties would probably cost more than paying a developer to do the work...
There are a number of approaches that can be taken to generating funding for fixing bugs of interest to the Mozilla user population.
The first is the model used by TransGaming for development of WineX. Basically people buy a subscription to the service. The subscribers then get to vote for what they want done next on a monthly basis. The top ranking games/features then are given a priority by TransGaming. See <http://www.transgaming.co…oting_reference_guide.php>
This answers the community desires for vote counts (previously bugzilla, but in this proposal a separate mozilla subscribers database) to actually mean something and it gives members of the community an incentive to support the Mozilla Foundation.
A second are the models similar to the efforts of the coSource or sourceXchange or the Free Software Bazaar. Each of these had slightly different mechanisms but the idea is that someone puts forward a proposal and developers can bid upon the work. This does have issues that would need to be worked out, but it is still early to give up on the idea completely. GNOME is experimenting with "bounties" to fund specific work items. See <http://www.gnome.org/bounties/>
As one of the leading open source projects, I would like to see the Mozilla Foundation try different options. Ideas that fail can be junked. Ideas that succeed (like Bugzilla and Bonsai) can then be picked up and used by other open source projects. I don't want the Mozilla Foundation to waste time but I do think that it should experiment with different ideas.
[Off Topic (OT): The automated use of multiple "Re: " prefixes by the MozillaZine software seems awkward. In general, there should be only one "Re: ".]