Printable Version of Topic

Posted by: Alison

It looks like Brion is moving on to another company, soon. He's announced it http://leuksman.com/log/ amongst other places. While I wish him well - he's one of the good guys - I have to wonder what this is going to do for Flagged Revisions and some of the other important software updates. Knowing geeks as I do, he'll probably still work voluntarily on the sidelines - I know I would.

Anyways - I guess Jimmy's going to be hiring programmers and CTOs again soon

Posted by: Kelly Martin

It will, no doubt, be used as yet another excuse to delay flagged revisions. There seems to be no end of reasons for not implementing them.

Posted by: Malleus

I doubt that Brion's departure will make much difference, or that it was triggered by the ongoing flagged revisions fiasco.

Posted by: Kelly Martin

Concur. I doubt very much it has anything to do with flagged revisions, or even that he's terribly much disenchanted with working there; instead, Evan offered him something better.

I suspect there will be long-term consequences for Wikimedia, but I expect they'll muddle through with perhaps some difficulty.

Posted by: MBisanz

I agree that I doubt it was triggered by flagged revs, but I would disagree that it will not make a difference, Brion has been with MediaWiki since it was created and is probably one of the only individuals out there with a full idea of what is actually in the software (undocumented extension, TS apps, external apps, etc) as well as what direction the software should be developed in (what things to include in core, how to make it more stable, more usable, things not to do with it, etc). I do have confidence is our other paid devs, but I will not undervalue Brion's immense role in the project.

Posted by: Milton Roe

Well, golly. I thought from his interviews that Jimbo coded up MediaWiki all by himself. I swear there were heroic images of him crawling though the server stacks personally, MySQL handbook in hand, to track down a bug. Think of Scotty crawling down a Jeffries Tube in the original Star Trek, to stick a magnetic probe up the antimatter stream.

Guess Brion has done as much MediaWiki coding as anybody, and is getting a bit miffed that all his efforts are going for other people's Wikiprojects that are either non-profit or run badly and low profit.

So, what's his big idea for the next internet thing? Well, he writes in the link you give:



Pegging my irony-meter, as that's not that much different from how Wikipedia screws people now on BLP. Heh. Except you can't use software to protect you from WP, as the other side has cyborgized by turning innocent brains into human writer-bots, to dig up personal dirt about you. That's hard to match.

And Brion promises he'll help keep WP doing it, at least for a while. Meanwhile, his own project to help people protect their "presence" on the web. From sales organizations with a lot fewer brains working for them than WP has. Due to the fact that for-profit organizations have to PAY most or all of their people, and don't have most of the work done by volunteers....

Yeah, I know. Brion can't be accountable for what they do with his software. But somehow I don't think he's quitting in moral outrage. Unless it's moral outrage at running WP's software for 4 years and not getting a piece of the pie, because there isn't any pie...

Posted by: dogbiscuit

Usually, if the software is dependent on some individual then it can be a good thing to get them to move on and knock it into shape.

However, what may be sound for a professional software company may not be the right approach from a bodge and dodge shop (and somehow, I can't imagine that Jimbo set up anything else). A bit less time on the trampolines, and bit more time documenting the bits that are too quirky for anyone's sanity.

Posted by: Kelly Martin

I think the one thing we can absolutely count on (we're already seeing it with http://lists.wikimedia.org/pipermail/foundation-l/2009-September/055389.html) is Erik heaping blame for any perceived failings of the tech team on Brion, even though the tech team is also Erik's responsibility.

Posted by: dogbiscuit

...and Brion, just to rub it in, after Erik whines about pressures, conflicts, lack of sympathy for the poor souls who are so stressed that they can barely find time to trampoline, and even Jimbo chimes in with some hopeless hand-wringing to boot

switches on flagged revisions on the test site.

JFDI

Posted by: thekohser

Yeah, right.

And http://flaggedrevs.labs.wikimedia.org/wiki/Special:UserRights/Thekohser.

Posted by: Alison

02:51, 29 September 2009 Gmaxwell (Talk | contribs | block) changed group membership for User:Thekohser from (none) to Administrators and Editors ‎ (GregoryCabal)

"GregoryCabal"

Posted by: One

Wouldn't an officer do that?


That does indeed seem pretty passive aggressive.

Posted by: MBisanz

That link only works if the viewer is also an admin, here is a link for the rest of us: http://flaggedrevs.labs.wikimedia.org/w/index.php?title=Special%3ALog&type=rights&user=&page=User%3AThekohser&year=&month=-1&hide_review_log=1

Posted by: anthony

"http://firstrayofsun.blogspot.com/2007/07/two-letters.html..."



He did just get demoted.

Posted by: thekohser

I feel sorry for these us admins who have to learn how to actually do these flagged protections. I can't http://www.wikipediareview.com/Image:Page_stabilization_shot_from_Flagged_Protection.jpg!

Posted by: Malleus

Exactly. Writing PhP "programs" doesn't seem that difficult to me, but the irony is that software shops generally can't afford outstandingly talented programmers (no idea whether Brion is in that category or not), because their work needs to be understood by those who are left behind when the prima donna moves on.

Posted by: anthony

Without outstandingly talented programmers, how is anything going to get done?

You just need to hire outstandingly talented managers as well, to make sure, among other things, that documentation gets written for the outstandingly talented programmers who take over to use to get up to speed.

Posted by: dogbiscuit

You need outstandingly talented people to have the great ideas - but generally it is best not to let them implement any more than you have to. Most companies I've worked at have had a prima donna expert, and most have been better off when they have caged them. Most bright ideas end up having massive hidden maintenance costs.

Posted by: anthony

That's an interesting position. Having always been on the "outstandingly talented" side of that equation, I really can't understand it. Nine times out of ten my great ideas don't get implemented if I'm not heavily involved with the implementation myself.

But if it works for you, I guess I can't blame you.

As for me, I suppose that's why I moved out of the software industry and became a tax accountant. It was far too frustrating to watch my creations destroyed by idiot code monkeys managed by idiot project managers. I guess I got spoiled in my first job, working on Unix kernel code with a whole team of outstandingly talented individuals, led by an outstandingly talented manager. In hindsight I should have never quit, but how was I to know that the rest of the world wasn't going to run that same way?

Posted by: Kelly Martin

This is why you often find the "founder" of many successful internet companies ensconced with a title like "Chief Scientist", usually with a nice office, but no direct reports. Such people are paid to Have Ideas and Convey Them To Others, who then turn them into viable product. Expecting people like that to also manage or direct is usually counterproductive.

Rumor has it that Cray paid people to steal his designs from him so that he wouldn't destroy them first; his rampant perfectionism would lead him to scrap perfectly viable ideas because, in the course of setting them down on paper, he'd realize some even better way to do something and would be tempted to throw the whole thing out. The key to his success in business was in not throwing out those "imperfect" ideas.

Posted by: CharlotteWebb

Gee I guess he's implying that the German, Polish, Hungarian, Russian, Arabic, Alemannic, Esperanto, Interlingua, and classical Chinese Wikipedias are not a good enough prototype.

Posted by: anthony

Okay, but what about the people that are outstandingly talented at implementation? What about the Wozniak's of the world?

Posted by: dogbiscuit

That's because my outstanding talents are in digging companies out of the mess outstandingly talented people have dug themselves into.

It does depend on what sector you are in, but the general drudge of moving data from keyboard to disk, across to another bit of disk and occasionally back to the screen rarely calls for talent, what is needed is blatantly obvious, comprehensible, maintainable code that everyone can agree what it does. That goes for clever SQL too - untangling dozens of subclauses of heavily nested Oracle SQL and having a few temporary tables instead works wonders for proving that the code works in all possible situations ("But that shouldn't be null so it's not my fault!??!").

I'm too scarred from impossible to follow C++ template code to consider clever coding to be a good thing.

Posted by: Kelly Martin

That's one of the things that struck me when reading through Mediawiki's code over the past year: there are a lot of cases where they've made "clever" use of various PHP array manipulations to make the code "more clever" that, in the end, impair both performance and readability. Using an O(n ln n) array function to replace a parallel loop (O(n)) is not "clever", even if it does mean fewer lines of code. Then again, if you are interested in efficiency, you probably shouldn't be coding PHP in the first place.

Posted by: anthony

So your problem isn't with outstandingly talented individuals, it's just with certain types of outstandingly talented individuals. I can agree with that.



There's no talent involved in producing "blatantly obvious, comprehensible, maintainable code that everyone can agree what it does"? If there's no talent involved, why don't we just get the computer to do the task itself?

Are you talking about a software company, or a non-software company which has custom-built software? There's a mountain of difference between the two. Your suggestions seem to be geared toward the latter.

Posted by: Friday

Yeah, there are a few niches where "clever" code is Good Thing, but for general purpose application development, easy-to-maintain code wins nearly every time.

If you're writing code that other people can't maintain, chances are you're not exceptionally talented, but rather an egotistical control freak.

Posted by: anthony

And it takes great talent to know where those "few niches" are, and where they are not.

And moreover, it takes a good amount of talent (maybe not great talent) just to write bug-free code outside those "few niches".

Given "dogbiscuit"'s explanation that his experience is "in digging companies out of the mess outstandingly talented people have dug themselves into", I think I see where he's coming from. But I'd venture a guess that in most situations (unless you're dealing with really simple software) if you had removed the talented individuals from the original implementation, the software would have never gotten written in the first place. I've never seen a group of mediocre programmers capable of producing a reasonably complicated piece of software all by themselves. And I've seen many groups of mediocre programmers try.

So I'm sticking with my original response. Yes, you do need outstandingly talented programmers to be involved in the implementation. But you *also* need outstandingly talented managers to ensure that they are creating a highly maintainable product. (In some cases you have one person that can handle both of these tasks, but in all but the smallest projects this is rare.)



Quite true.



So, to get back on topic, in all likelihood Brion is a great programmer. But in my opinion he was a terrible CTO.

Posted by: Somey

This is all true, but even the most "talented" managers will always tell you that they can't succeed without plenty of time, adequate funding (which really means staffing), and upper-level management support. (Some degree of organizational autonomy is also nice, though arguably non-essential.) If there's a customer involved, it also helps to have a good liaison person with the customer - I myself have been burned more than once by that problem.

IMO, those things are quite often what separates the "talented" managers from the untalented ones. Managing projects is fairly easy when you have the right infrastructure in place, particularly people who are willing to do what they're asked to do, without your having to yell at them or threaten to fire them.


So will they hire an actual CTO to replace him? That would probably cost more than hiring someone out of the WP/MediaWiki "community" who has little management experience and just needs a better job. I'd be surprised if they do that, quite frankly.

Posted by: dogbiscuit

Oh, you don't try and get away with mediocre, you need good, just not God-like. Basic professionalism.

Posted by: anthony

I thought that was http://www.mail-archive.com/foundation-l@lists.wikimedia.org/msg06459.html.

Whoever they choose will still be under Sue and/or Erik, though, so unless they're really good *and* really bold, not much is likely to change.

I'd expect someone at about the competence level of Sue, which would be somewhere in between someone out of the community and someone who can do things right.

Posted by: Kelly Martin

Concur. The skills that make someone an excellent programmer do not in any way translate to being an excellent CTO. Someone who excels at both does so by dint of being skilled at two different things, and one doesn't really predict the other (except insomuch as general intelligence is a baseline requirement for either).

I made a bid for the Wikimedia CTO position back in 2006/7 (before I realized the depths of Jimbo's dislike for me, and also Erik's), and it was hard for people in the WMF to wrap their heads around the idea that Brion could be an absolute genius with PHP and Squid and such and yet be subject to criticism for his poor handling of things like budget and manpower management. As an example, at that time, mainly because of lazy negotiating, they were paying at least 80% more for IP transit than they should have been. Given that at the time, IP transit were something like 25% of the budget, this is pretty much unacceptable incompetence.

Then again, this is the organization that used the services of a biologist as its CFO for over two years.

No, they'll hire some old crony of Sue's (or maybe Erik's) who has something resembling IT experience and is in need of a job. And who, in all likelihood, will be Canadian.

Posted by: anthony

Is that who it was who took a linear plot of year-by-year hardware *purchases* (not even factoring in useful life) to guesstimate future hardware needs? I remember arguing with whoever that was.



LOL. I don't know if that is more perfectly hilarious or more perfectly true, but it certainly has aspects of both.

Posted by: Milton Roe

Or you have poor communications skills, which is far more common. Clear English annotation about "what I'm doing right here" can save anything, but you can't always get the coder to write it in any way that can be understood when he or she is not there. This is really the same technical writing problem as in any technical documentation. The expert who is doing the job is not necessarily the best person to explain it to others, but coding is one place where you need to, right at the same time, or else.

Posted by: dogbiscuit

MediaWiki has the classic syndrome of copious comments for the obvious - when things are going well and life is but a cut'n'paste - and then when things get tricky, we are left with "Don't change those double quotes back to single quotes because it breaks things" (with a few other "I don't know when this cariable gets set, so I've got to have this bit of code here").

Good coding needs very little comments because you use long, awkward, unambiguous self-explanatory variables and procedure names, and rely on the editor to do the typing for you (Visual Studio is a dream).

I lived through the James Martin Workbench phase, where it became abundantly apparent that if you wanted code, you needed to write code, and all the bluff about turning analysis into code simply meant you had highly paid analysts writing code (or worse, end users*).


*At a Leeds hosptital, IBM were facing Oracle. IBM were selling their good ol' American package, not a great fit but good enough; Oracle were selling a database, plus a load of ticks in the boxes of "can it do this that and the other?" because, as a database it could be programmed to do anything - in this case, the ultimate healthcare "expert system". I pointed out that Oracle were making nurses and doctors programmers, and the lack of wisdom of this, when they had real jobs to do - and they were not trained to recognise the implications of coding up a system.