We're having some work done on our condominium complex and a month ago the construction company posted the first draft of the schedule, a nice little Gannt chart that says the first thing they'll do is have external catwalks installed by June 7. And here it is, June 19, and...no catwalks. It reminds me of the very first time I tried to use scheduling software, Symantec Schedule 1.0 I think it was. I made a nice little Gannt chart for *Gryphon Masters* and then watched in horror as the days went by and that very first bar got longer and longer. As always, it's good to know that construction companies have just as much scheduling difficulty as software companies.
Now, when I saw that Gannt chart the first time I was really excited. Things are finally moving, I thought. We'll have these sagging walkways with their dryrot replaced in no time. I mean, there's a Gannt chart! Which is the nice thing about Gannt charts. They make everybody feel really good for about a week.
And then you hit that first milestone (or inchpebble) and the task isn't complete and you wait and you wait and you realize the Gannt chart is just words, words, words. Even if you update the Gannt chart every day it's depressing: the day before yesterday we realized we'd actually be done September 19th. Yesterday we realized we'd actually be done September 20th. Today we realized we'll actually be done September 21st. Wait a minute...if this keeps up we'll never finish!
Which is about the time I throw out the Gannt chart, which is too bad, because in theory, in a couple of months, things will actually settle down, and the Gannt chart will start looking like an approximation of reality. I just have to learn to get over that hump.
Post a comment
Your Information
(Name is required. Email address will not be displayed with the comment.)
Remember, when they finally get the catwalks up that the delay is not just from June 7th to when they get the catwalks up, you have to assume that a similar percentage of slippage will occur on each of their other tasks. This will give you the pessimistic end date of when your apts will be looking good again. I would then add 3 months to that date, since contractors are notoriously LATE.
It's not just games that have a bitch of a time getting things done on time, you either have people who have learned how to estimate times and schedule or you don't. Experience helps a ton, as does thinking about this sort of thing, as does having more than one person in your company (like many contractors are). Complexity of the job makes things much harder, as does unmovable ship dates, as does fluctuating team composition, and most certainly a moving target of what you are making.
Posted by: Chris Busse | June 21, 2004 at 01:28 AM
More Gannt Chart thoughts...
Making a beautiful Gannt chart IS very useful at certain times throughout the project, especially the beginning. It's like making a GDD. The process of thinking through each item and trying to relate it to all of the other tasks that need to be done is a must do for all projects taht expect to ship anywhere close to on schedule.
Maintaining an up-to-date daily gannt chart, however, is a collosal waste of time.
Tracking progress on tasks and people is essential, seeing that you've slipped one day, then another, then another each day doesn't help.
It's like getting on the scale each morning when you are trying to lose weight. There are so many factors that determine weightloss, that looking at it on a daily basis is just a bad bad idea. Most doctors and weight people suggest not weighing yourself but once per week. It takes the perspective of some time to be able to rationally tell if what you are doing is working or not.
Game dev is the same way, daily is just plain ridiculous, not only of the amount of effort it takes to do this sort of thing, but in the miopic results that you are seeing on a daily basis gives you the wrong information (you don't see the forest for the trees). I think anywhere from semi-monthly to bi-monthly is the right time frame, depending upon the length of the project and the phase of the project you are in. Any longer and you run the risk of being WAY off course, any shorter and you end up shifting so often that no one can hit their stride. I'm convinced that when devs are allowed to progress "naturally" on a task or series of tasks, that's when the best work gets done.
Plus, then you can save the shifts and this must get done nows for when you really need them; and the team will respond better in those situations if they haven't been told wolf 1000 times already.
Posted by: Chris Busse | June 21, 2004 at 01:42 AM
Chris - Agreed.
Jamie - Perhaps instead, we should all be worried that our Gannt Charts approximate those of the construction industry? :)
Posted by: Scott Macmillan | June 23, 2004 at 07:36 AM
Having just finished a two month course in which we made a 'thesis game', daily Gantt updates pretty much made me feel like I'd been bitch-slapped daily. There was tons of, "Oh man, the shotgun was supposed to be done two days ago! Okay, the sniper rifle's cut!" in my group.
Of course I'm sure our quickened schedule made things different than a normal game development cycle, but I despised the Gantt chart none-the-less. I think if we did it weekly it would've been better. I could definitely understand monthly or bi-monthly working better in a real cycle.
Posted by: Jeffool | June 23, 2004 at 05:21 PM
But wait; if you hadn't had the Gannt chart, you wouldn't have realized the sniper rifle needed cutting, and the overall quality would have suffered, right? So yeah. Forget daily.
Posted by: Jamie Fristrom | June 24, 2004 at 08:00 PM
I think the Gantt chart is helpful and a good thing to have, but I just meant to rail against daily updates.
(This is what you get when don't preview, odd paragraph structure. That and the urge to tell everyone it's spelled 'Gantt' because you're a grammatical ass. ;) )
Posted by: Jeffool | June 25, 2004 at 10:19 AM
You know, I was going to spell it that way, then went with the crowd. :D
Posted by: Scott Macmillan | June 25, 2004 at 11:12 AM
Actually, I almost did as well. The power of 'the crowd', huh? Honestly, I Googled it to make sure I hadn't been spelling it wrong for the past few months.
Posted by: Jeffool | June 25, 2004 at 04:23 PM
I just looked at Jamie's post and aped.
Posted by: Chris Busse | June 26, 2004 at 01:00 PM
How do you make everyone in the team refers to their respective Gantt Charts and at least know that they are behind/ahead schedules?
It's great to tell the team,"Hey, you are way ahead of schedule! We can really push this game out sooner than we thought!".
And it's another to say," Hey look, get your act together okie?!"
Maybe it's still back to old school voice communication...
Posted by: noobee | July 03, 2004 at 09:12 AM
You know, I didn't realize I spelled it wrong until today - on the bright side, this post is the third google hit for anyone else who misspells it.
Posted by: Jamie Fristrom | August 20, 2007 at 12:03 AM
Salt, meet wound. I think it was just a month or so after this I completely screwed up a chance at a job that someone went out of their way to set up for me. How? Sent in a short written piece that, even if the idea behind it was genius, was so horribly written and spelled that they couldn't have hired me if they wanted to.
The shame was heavy I tell you. My tail was tucked so far between my legs it looked like a second, yeah, well, y'know.
Posted by: Jeffool | August 20, 2007 at 09:24 AM