Slashdot Mirror
Microsoft's Not So Happy Family
D.A. Zollinger writes "Reports from Redmond are that Microsoft Employees are not happy with the double delay of Windows and Office being pushed back into 2007. EETimes is reporting that some Microsoft employees are calling for the termination of several top managers Including Brian Valentine, Jim Allchin, and Steve Ballmer for the delay debacle. The report references a blog by Who da'Punk, an anonymous Microsoft employee who asks, where's the accountability for failure? So far the blog entry has generated over 350 comments from Microsoft insiders and outsiders."
Microsoft has a bad reputation with regard to the quality of their code. But they have a really good reputation for shipping products.
This is news to me. Maybe you mean eventually shipping product, but their general reputation is for always being years late and always dropping features to make even the late dates.
Infuriate left and right
I found the following post (on the MS Blog comments) by someone who is probably a MS employee.
----snip-------
Talk around the vending machines in legal is that the delay has nothing to do with coding, slipped schedules or anything else. That's why very few heads will actually roll and most will simply shuffle positions. Actual reasons have to do with no product, NONE, shipping until after the mess with the EU is cleaned up. From what I've heard so far, if there are further major delays with EU that can't be solved by set-asides and scholarships, then expect another major delay beyond what has already been announced. At 25-40% annual compounded growth rates for Linux servers, the last thing that's going to happen is for the EU to be able to do what US-Justice failed to do, which is force disclosure of MS server protocols so competitors can copy MS's IP and gain market share in the market segment on MS's dime. Samba has never been 100% compatible and that's the way its going to stay, come hell or high water. Regardless of how much time/delay it takes, Samba and Vista will never be as interoperable as Samba is with PDC, AD, AS currently. If it takes another 6 month delay, another 9 months, whatever. Eventually EU will capitulate whether Commerce and the WTO has to step in or not. Server space market share has either reached a tipping point, or already passed a tipping point depending on which internal study you read. Whichever study you read/believe, make sure its one of the ones that takes into account free installs of their versions of AS/ES, such as Cent/OS. According to those studies, the server space has already passed the tipping point, that's why we're seeing what's happening with Mass/ODF/XML, and some of the large desktop migrations that have been documented internally. And remember, any large migrations you get a whiff of, you know where to report them, get details and do it. A single 6 digit desktop migration has repercussions far and wide on many other customers and partners (and media), and we are staring at over a dozen of them and have been unsuccessful in turning any of them around so far.
So unless anything settles with the EU in the coming months, expect further delays regardless of what they are blamed on.
--------snip--------------
Software most certainly falls into the category as well. Any process where an advancement is used to produce further advancements gets an exponential nature.
I'm struggling to think of any advancement (at least in recorded history) that *doesn't* build on prior advancements.
In software it couldn't be more clearer. After you write your first compiler in machine code, writing your next compiler will be much easier as you base it on the previous step.
Right. But that doesn't mean the 6th revision of that compiler will be as quick to develop as the second.
Indeed, based on the history of software development thus far, the chances of it taking anything less than an order of magnitude *more* time to develop are quite small.
When they started the development of Vista, they had already an operating system to build on and a variety of advanced development tools. With that as a starting point it should have been an order of magnitude faster than the previous step.
Your theory sounds nice, but I'm not aware of any mature software projects for which it has actually happened. In pretty much every case, the more developed a codebase is, the *longer* it takes for subsequent versions to appear (assuming the changes are on the same scale).
I think it would have been perfectly reasonable for Vista to have taken on the order of 3 - 4 years to develop (in line with Win2k from NT4). In fact, if you take into account that they basically "started from scratch" again around 2003, that's about how long it *will* have taken. The real reason Vista (NT 6.0) is late is because of the "lost" 2 years of work between XP (NT 5.1) and Windows 2003 (NT 5.2). Arguably, Microsoft should have released an "XP second edition" (NT 5.3) in 2003/4 - but since the obvious differences between it and XP wouldn't have been large, it was probably considered a waste of time.
Basically, the recent NT family tree looks like this (it's rather difficult to do ASCII art on Slashdot, I hope you can understand):
Windows 2000 (NT 5.0)
..............V
............Windows XP (NT 5.1)
.............V..............V
Windows 2003 (NT 5.2).....Longhorn (NT 6.0)
...V.......................V
...V..........(Code discarded)
...V.............V
Windows Vista (NT 6.0)
Basically, XP branched off into Windows 2003 and Vista (then Longhorn). But around the time Windows 2003 was released, they decided that it was a much better codebase to develop Vista from, so the existing Vista codebase was scrapped and the project started afresh from Windows 2003 (more accurately, a lot of the "Vista" development and "Windows 2003" development was the same and, technically, kept).
It's interesting to note FreeBSD had similar problems around their 4.x, 5.x and 6.x codebases. Arguably, the VM fiasco in the early 2.6 kernels was along similar principles, if not scale (but then again, the Linux kernel is a dramatically smaller project than Windows, so in relative scale they might be somewhat comparable).
The point here is that software development is not a field where advancement is anything close to "exponetial". If anything, it's the exact opposite - the more mature a codebase gets, the *slower* releases become. The "software development" curve looks more like a bell, than anything linear or exponential.