Yes, you can build lots of stuff quickly. But that doesn't mean it is production ready. The devil is always in the details.
The beginning of any project is fast, you always thing 'wow, this is going just great, at this rate I'll be done next week'. And then it starts to slide little by little you slow down, no matter how polished your dev environment, that's just a matter of degree.
So, after a month has passed and you start wondering what happened to make you slow down you start to realize that this project is probably going to be just like all the others. Once you've implemented your days list of fixes and features you find yourself with a longer list for the next day.
And that's before you let the unwashed masses in the front door, that's when the fun really starts.
I'm not sure that an opensource version of StackOverflow would be as useful as StackExchange. Let's ignore the re-implementation issue, and assume that an open-source clone appears by magic tomorrow morning. You still have the support problem.
Anyone doing this as more than a hobby needs a supported solution. The reason to pay $xxx a month is so that you can outsource worrying about the site's maintenance and upkeep. The cost (in man-hours) of building such a clone would be dwarfed by the hours spent bug-fixing and performance-tuning and dealing with hardware/software issues. If the site is ancillary to your business (if it's your customer support site, for instance), then there's a lot of sense in letting someone else handle the support, especially if web software isn't your core-competency. At $1300 for a dedicated, hosted server, you're essentially betting that it'll require more than 12 programmer-hours/week ($15.6k/year hosted vs. $50k/year for the programmer) to maintain the open-source version and run the hardware. And you're betting that someone at Fog Creek dedicated to maintaining Stack Exchange is going to be better at it than whomever you lasso into maintaining it part-time.
3 comments
[ 2.6 ms ] story [ 14.3 ms ] threadYes, you can build lots of stuff quickly. But that doesn't mean it is production ready. The devil is always in the details.
The beginning of any project is fast, you always thing 'wow, this is going just great, at this rate I'll be done next week'. And then it starts to slide little by little you slow down, no matter how polished your dev environment, that's just a matter of degree.
So, after a month has passed and you start wondering what happened to make you slow down you start to realize that this project is probably going to be just like all the others. Once you've implemented your days list of fixes and features you find yourself with a longer list for the next day.
And that's before you let the unwashed masses in the front door, that's when the fun really starts.
You can put away your tin foil hat now.
Anyone doing this as more than a hobby needs a supported solution. The reason to pay $xxx a month is so that you can outsource worrying about the site's maintenance and upkeep. The cost (in man-hours) of building such a clone would be dwarfed by the hours spent bug-fixing and performance-tuning and dealing with hardware/software issues. If the site is ancillary to your business (if it's your customer support site, for instance), then there's a lot of sense in letting someone else handle the support, especially if web software isn't your core-competency. At $1300 for a dedicated, hosted server, you're essentially betting that it'll require more than 12 programmer-hours/week ($15.6k/year hosted vs. $50k/year for the programmer) to maintain the open-source version and run the hardware. And you're betting that someone at Fog Creek dedicated to maintaining Stack Exchange is going to be better at it than whomever you lasso into maintaining it part-time.