We recently saw the news that the German city of Freiburg had decided to end its open source migration and instead switch to using Microsoft products again. The rationale provided seemed curious to me - after all, at the same time the German city of Munich announced total savings amounting to ‚¬10 million from its own successful and ongoing migration.
What seemed odd was there was no account of how they changed course to make the migration succeed. Munich learned lessons from early challenges and updated its strategy in order to succeed. But not Freiburg.
From what I could see, instead of ditching the old versions of MS Office and OpenOffice.org they'd started with and installing up-to-date LibreOffice using expert in-house help, they had just hung on to outdated software and expected staff to muddle through to success. When that didn't happen, they blamed the software and not the strategy. Everything was in German, so rather than risk misinterpretation I turned to German-speaking friends in the technology industry to explain the report to me (if I got anything wrong, please tell me - the documents seemed very complicated).
My (guided) reading shows three points of concern in the situation over the last four years. First, the only ongoing expenditure in support of the migration is running costs of less than ‚¬15 per seat per annum, all associated with licensing supposedly superceded proprietary software. Second, substantial one-off costs of around ‚¬231/seat associated with interoperability - a topic that is always an indicator that proprietary software is controlling people’s thinking. Third, no obvious investment in ongoing community engagement or equivalent commercial subscriptions for open source.
First let's consider what Freiburg were buying. Most of the costs appear to be one-off, suggesting that cost-cutting was the dominant focus of the strategy. A translation of the German materials shows these one-off costs:
- ‚¬25K training
- ‚¬55K licensing a legacy commercial product for "interoperability"
- ‚¬55K consulting cost for the migration (FormularMax)
- ‚¬20K consulting cost for the migration (macros)
- ‚¬6K developer cost for migration (Calc)
- ‚¬60K introduction of Wollmux
- ‚¬240K personal cost project management (‚¬60K 80% committed to the project over 5 years)
That comes to ‚¬461K in total - around ‚¬231 per seat.
Compare this with the city of Munich, which reports total migration costs of ‚¬270K total for migrating seven times as many seats to both Linux and LibreOffice (not just OpenOffice.org on Windows like Freiburg) - a migration cost of ‚¬18 per seat.
Given that, and the fact that Freiburg's strategy still retained Microsoft Office for some tasks, one has to wonder how much of these 'migration costs' were actually maintenance costs to keep Microsoft Office afloat in the existing workflows. Indeed, the only reported annual running costs are for a legacy version of MS Office at ‚¬30K total per annum - about ‚¬15 per seat.
The Freiburg report suggests that its new solution of buying Microsoft Office enterprise licenses would cost ‚¬60 to ‚¬80 per annum, a rise of at least ‚¬45. Even going for the smallest price on offer would at least quadruple the current annual running costs. I'm using the lowest numbers possible there - there are other, higher numbers quoted in later additions made at the request of the Mayor's office which would make it even more expensive.
Increased license costs are one thing. But the report is mostly silent on the one-off costs of such a reverse migration. All of the above migrations have to be performed again, backwards, without a budget, thus proving that the administration is still in the trial and error mode it claims to have ended. Since the UI of newer MS Office versions is radically different from older versions it would be foolish to assume the training costs would be smaller. It can therefore safely be assumed that the one-off costs for this migration are at least just as high as the costs already incurred.
When it comes to interoperability, the report claims additional running costs, raising the annual total to ‚¬280-‚¬380 per seat. It adds in to the annual budget:
- ‚¬24K personal cost for centralized IT
- ‚¬500K for interop issues (‚¬250/seat)
The first question it immediately raises is: If the city administration was indeed in such dire straits, why didn’t it act upon it and spend serious money on a service contract to deal with the interoperability issues? The recent OSB Alliance deal showed that even smaller amounts can go a long way. And -- assuming the numbers are true -- it would have paid off after just one year.
An alternative would have been to hire a team of 3-4 excellent software developers locally, reporting to the city administration alone and fixing their issues. Maybe that would have cost as much as ‚¬500k -- but as the city was advised to spend the same amount per annum on Office 365 subscriptions by the consultant in the report, this does not seem an issue.
So just from a local point of view the city just decided to go for a solution that according to the report:
- creates additional one-off costs the same as those just paid for the migration to open source
- saves taxpayer money by firing 8-9 local workers no longer required because of easy interop
- spends that saving to pay a foreign company for license fees each year
- having the long term strategic advantage of being independent by using open source
- providing 3-4 additional well-paid local jobs with the same implications on the city budget and output of the city administration
- feeding the local economy a large slice of the money invested.
It's complicated, and took a while to study, but I don't think the numbers really add up. It seems Freiburg has not invested in its open source solution in any way likely to make it succeed, but has rather left it to fail “to save money” and then when it has, blamed the open source software instead of the flawed strategy. The structure of the report makes this look a conscious frame.
It claims high cost estimates for the open source approach, attributing every possible cost to it, while on the other hand only naming the costs for the Microsoft solution where they would be easy to research and then claiming all other costs as marginal. The report seems to avoid providing any real basis for comparison -- not even estimates -- and obfuscates by never clearly separating running costs from one-off costs. While lengthy and complex, the report is missing many of the interesting numbers, making it unsafe as the basis for any concrete decisions without further research
That the council still did so is the real tragedy. As Timothy Simms of Germany's Green Party implies, the recent change seems to have at its heart an attempt to politically justify a return to a preferred supplier while resisting a successful move to open source. If that is indeed the case, the lesson Freiburg provides Europe is that all solutions require investment, and strategies that fail to do so guarantee failure, regardless of the technology.