So after a few years of reading thedailywtf I'm at the point of contributing something.

The setting is the school system in the capital of a important european country.

After a few decades of "Do-Your-Own-Network" and "Use-The-Software-You-Want" in all city schools, the local government decided to commence a project to standardize hard- and software. It the beginning, it sounded fairly good - all schools should (at last) get state-of-the-art PCs, servers and networking components, they got a couple of headmasters together to ask them what they needed and how they were doing their
stuff and took all these specifications together to get a bidding from different software manufacturers.

They didn't ask less then a system that could manage all student data beginning from name and address, going on with their school career, all grades, reports, forms etc. to calculating the degree grades in the end, course scheduling, teacher scheduling and so on.

They expected some big-shot companies to bid. Then they found out, that none such software existed, that would fulfill all specifications. Solution: they bought TWO DIFFERENT programs, one of which would have to be heavily adapted. One to do the database stuff (student names, grades, passes and fails, course enrollments, teacher names and addresses, forms and reports - in the following named "Columbus") and ANOTHER ONE, which should do the scheduling based on the data of the first one
(which is named let's say "Schedulis"). Well, Schedulis has been on the scene for 20 years, good piece of software for managing timetables / schedules, but not good for managing student data.

Footnote: "Good Old Software", which most of the capitals schools are using wasn't part of the bidding process, because the company was to small (1-2 Programmers). Well, as it turned out, the company selling Columbus has got ... two programmers.

Small problem with selecting two different software packages:
the two use entirely different data structures. No problem: data can be transferred between them via ... Tada! ASCII Files.

School administration now is really far easier than before.
Let's say, one of the school administrators wants to plan the schedule for the next year. With the new solution we need only a couple of easy steps.

1. Secondary school administrator (SSA) takes student selections and determines, how many courses will be needed (Columbus)
2. Headmaster says which teacher will take which courses / classes (wooden table)
3. SSA inputs data into Columbus and exports ASCII File.
4. Deputy Headmaster (DH, who is responsible for the schedules) imports data into Schedulis and runs an optimization to assign students to courses based on all their course selections.
5. DH exports data (ASCII).
6. SSA imports data in Columbus and sees courses, students and teachers.
7. DH plans rest of schedule in Schedulis.

If something needs to be changed in both databases (Teacher doing the course, student taking the course), please change in one of the two databases AND RE-EXPORT / RE-IMPORT. *sarcasm*: No we don't think that any problems doing this a few times...

That's what we need new software for. The "Good Old Software" we still use does most of this in one environment, even when two SSAs were planning to different student years at the same time... and by the way Good Old Software also has got all the reports we need and prints them nicely. In the training, one of the trainers told us: "If you can, get a
copy of Crystal Reports yourself. The software will not be able to print all the reports you need in the beginning - and if you wait for The-IT-People-Above to fix the problem (e.g. create the report), school year might be over without printing the form / report / etc. you would have needed.

Next bug: in our training, after test-running the process explained above, we tried to print a report in Schedulis hich should give us a list of all the names of students taking a course. What did we see? Only Columbus IDs. WTF?

Again during training, one of our SSAs also tried to input all the selected courses into columbus. Students should only have 32 courses. Does Columbus complain if you input the 33rd? No, it doesn't even have a label to display the number of selected courses.

I really didn't think you could mess things up THIS much, but at least I have something for the CS students I teach at my local school - every week a new story "How Not To Do Things - Part N+1".

knirf:

every week a new story "How Not To Do Things - Part N+1".

Failures I see in the process as described:

1. Asking principals what they want. A school principal typically doesn't have time to interact much with the school's admin software. That's usually the job of the school secretary and/or bursar.

2. Putting out a request for bids without first finding out what was already in use and whether any of that was already working satisfactorily.

Both of these are sadly familiar to me. We've only just managed to kick a similar mess in the head after six years of suffering and cut over to the reporting/attendance module offered by the folks who made our existing main school admin package. There have been a few teething problems but even so it's already far less painful than what we had before.

Schools are not known for deep IT budgets, so it takes a special breed of company to go into the business of writing software for them.  The two most prevalent models are "a couple of people passionate about making education better" and "heartless corporation that wants to develop a single product and then sell it year after year until it stops turning an annual profit"

I used to work for a company that invested no small amount of effort attempting to get educational software to run in modern environments (well, the environment was modern at the time, it is a touch dated now).  The typical story was that the developers had all been laid off and all new development was outsourced.  The only staff you could reach was the support team (also outsourced most times).  They had no idea how the software worked, nor did the outsourced developers.  If you could figure it out, they would love to hear the solution because it is a very common request.  On occasion, we managed to figure it out.  We didn't share.

Last I heard, SIF (http://en.wikipedia.org/wiki/Schools_Interoperability_Framework) was winning the war of providing for interoperability of school software (which, yes, typically only supports export/import to CSV) but high costs were keeping adoption low.  Not only was the server software outrageously expensive (and typically licensed per-student per-year) but there was no incentive for companies to invest in making their software SIF compatible--their customers were so heavily invested in the software after decades of use that brand loyalty would trump innovation.

Then theres things like the patent Blackboard received for e-learning.  Thats a fun read (I don't know the patent number offhand, and I don't know if it has been overturned yet)

At the time, I thought that the world of educational software was about as FUBARed as you could get.  That was before I started working in the defense industry.

knirf:

Small problem with selecting two different software packages:
the two use entirely different data structures. No problem: data can be transferred between them via ... Tada! ASCII Files.

flabdablet:

Failures I see in the process as described:

Ditto that.

Poor stakeholder analysis, leading to poor requirements-gathering.

Not understanding the problem = not producing a satisfactory solution.

Sack the project manager!

Kaosadvokit:

Schools are not known for deep IT budgets...

Amen to that. My company is one that services higher educational institutions and half of our clients make up incredibly poor excuses as to why they will not pay off their full bill.