by NoOneImportant » Tue Oct 15, 2013 5:36 am
Suggest everyone reread John's post... especially the bridge analogy. In large system development where development is rushed to meet unrealistic roll-out dates, internal development discipline disappears... what you end up with is this undocumented monster where nothing appears to work as planned. As the roll-out date approaches ever closer, everyone feels the pressure, and panic sets in; developers begin doing whatever they think they need to do, wear-ever they want to do it, and abandon all development discipline. The result: errors to appear to happen "at random" and leave no, or little trail as to who the offending party - software module - was who caused the problem - thus, as Gerald noted above, nothing appears to be certain, and thus every one becomes gun shy of using any output of any sort, as nothing can be relied upon.
Scalability alone is an enormous issue, what may work for 4,5, 10, 100, or ever a thousand users may not work when you have 10, 20, or 30 thousand users banging away on it. Time per user becomes super critical. If each user requires .3 seconds to do something, that's not an issue for a single user, but start stacking up individual users and it very quickly, with a relatively small number of users, becomes an enormous problem as the time may be cumulative. Yes it only takes .3 seconds for any given user but if you have 10 users that 3 seconds, 100 users that's 30 seconds and so on. As noted in John's article inexperienced programmers often don't understand the physical ramifications of what appears to them to be a simple programming request. All of these things come out in testing, and are addressed, if sufficient time is allowed to test things.
But again, this is simply yet another area of which the President is just grossly ignorant. He uses his PC, he surfs the Internet, and believes that he can command his heart's desire - after all, Microsoft can make Windows work, can't they? He has no memory of the early versions of Windows that were simply awful. He routinely cruses the Internet, he sees that it works, he can view Fox News any time he desires, right? He has no memory of the Internet's development period from the mid 70s to the mid 90s, times when maybe the Internet worked, and maybe it didn't. So, he rationalizes, why shouldn't Obamacare be used to orchestrate 300 million people into nationalized healthcare? So he gives the developers a couple of years - ... anybody should be able to write whatever software they need in a couple of years, right? Sorry, that's just not how it works - you design big, start small, then scale it up; fix the problems; then scale it up; fix the problems then scale it up. No one in their right mind goes from zero to 300 million users in one fell swoop - with the possible exception of Obama. The problem is much akin to fusion power generation. We know that it works; we arise daily to view a clear example in the sky; we know that it works in large bombs, yet getting it to work on a contained manageable scale isn't all that easy - perhaps Obama should simply command that controlled fusion simply work, because he knows just about as much about fusion as he does about large system software development.
Suggest everyone reread John's post... especially the bridge analogy. In large system development where development is rushed to meet unrealistic roll-out dates, internal development discipline disappears... what you end up with is this undocumented monster where nothing appears to work as planned. As the roll-out date approaches ever closer, everyone feels the pressure, and panic sets in; developers begin doing whatever they think they need to do, wear-ever they want to do it, and abandon all development discipline. The result: errors to appear to happen "at random" and leave no, or little trail as to who the offending party - software module - was who caused the problem - thus, as Gerald noted above, nothing appears to be certain, and thus every one becomes gun shy of using any output of any sort, as nothing can be relied upon.
Scalability alone is an enormous issue, what may work for 4,5, 10, 100, or ever a thousand users may not work when you have 10, 20, or 30 thousand users banging away on it. Time per user becomes super critical. If each user requires .3 seconds to do something, that's not an issue for a single user, but start stacking up individual users and it very quickly, with a relatively small number of users, becomes an enormous problem as the time may be cumulative. Yes it only takes .3 seconds for any given user but if you have 10 users that 3 seconds, 100 users that's 30 seconds and so on. As noted in John's article inexperienced programmers often don't understand the physical ramifications of what appears to them to be a simple programming request. All of these things come out in testing, and are addressed, if sufficient time is allowed to test things.
But again, this is simply yet another area of which the President is just grossly ignorant. He uses his PC, he surfs the Internet, and believes that he can command his heart's desire - after all, Microsoft can make Windows work, can't they? He has no memory of the early versions of Windows that were simply awful. He routinely cruses the Internet, he sees that it works, he can view Fox News any time he desires, right? He has no memory of the Internet's development period from the mid 70s to the mid 90s, times when maybe the Internet worked, and maybe it didn't. So, he rationalizes, why shouldn't Obamacare be used to orchestrate 300 million people into nationalized healthcare? So he gives the developers a couple of years - ... anybody should be able to write whatever software they need in a couple of years, right? Sorry, that's just not how it works - you design big, start small, then scale it up; fix the problems; then scale it up; fix the problems then scale it up. No one in their right mind goes from zero to 300 million users in one fell swoop - with the possible exception of Obama. The problem is much akin to fusion power generation. We know that it works; we arise daily to view a clear example in the sky; we know that it works in large bombs, yet getting it to work on a contained manageable scale isn't all that easy - perhaps Obama should simply command that controlled fusion simply work, because he knows just about as much about fusion as he does about large system software development.