Having worked for the French state and wrestled a few times with its IT services, I can tell you that the reason for choosing Microsoft isn't cost, or "efficiency".
It's that they only know Microsoft, they don't want to learn something else, and if there's a problem, it's Microsoft's fault, no theirs, so they don't have to deal with their own incompetence.
If you want an anecdote, we were working with SAS, a statistical software which required costly licences (more than a million € for a few dozens of workers). I suggested to switch to R or Python to the top director, who agreed.
First meeting with the service in charge, the chief opens with "ok, we are asked to change, but the goal here is to show that we tried, and found that it's not possible."
I resigned a few months after, as everything was in the same vein.
Sounds like Germany, and it's not just public services.
I used to work part-time in 1st level IT support in a local hospital when I was younger.
The main "theme" of my superior's work subjects there (2009-2016) was the migration from XP to 7. You heard that right.
And apart from the usual Office- and AD-Lock-In, the most problematic workstations of course were always ones with very specialized software. Virtualization and terminal services were in use, but the whole selling point of Windows was mostly put ad absurdum already, because they needed Windows licenses for dedicated machines running e.g. specialized MRT software, but those weren't even part of the main network anyway. They needed arcane syncing procedures anyway and Windows provided no value whatsoever on these devices. Same for the patient monitoring systems on ICU beds. These were using some "embedded" Windows and were rarely working in a stable way at all, nor way they connected to the networks running AD or the CIS (edit: seems it's called HIS in English)
CAD and stuff in the office divisions was similar, but with less integration needs (apart from network printing)
What I'm trying to say is: like in many offices, any slight change made users hostile, updates cost obscene amounts of work and money, and Windows didn't provide much more value compared to SAMBA. That is dated experience, I know.
But MS has not shown to be a trustworthy company in any of my work experience so far.
It was impossible to create working solutions without MS, yes, but the reasons for that never seemed to be grounded in actual value provided by an MS-centric software and networking structure.
It was just the one available commercial solution with enough adoption, and MS has been milking their target markets with these strategies for a very long time.
Making themselves "indispensable", even in machines where their software was used to run a terminal server, basically.
Hell, in my town, 3 years ago, they started to replace subway train LED signals with crappy Windows-CE-based software.
The effects are still noticeable... the whole infrastructure is still 80% worse compared to 10 years ago.
You recognize the useless Windows licenses by the occasional Desktop (seriously, google "cologne KVB windows trashcan"....), 90deg-tilted display, and of course 20% of the signage is out of operation on average now.
>What I'm trying to say is: like in many offices, any slight change made users hostile,
I remember doing a Windows 7 deployment for a customer, and one of their staff had 3 A4 binders. Probably 1000 pages, of typed notes about how Windows XP works.
The second she sat down at her new Windows 7 machine it was like she had lost her memory. She was instantly aggravated and couldnt find the start menu. Dropped her windows xp binders in the bin and told me to effectively "get lost" so she can relearn it from scratch.
I was thrown. Like its basically the same system with some UI tweaks but she was acting like I had pushed her onto debian or something.
I’ve had a few family relations approach technology this way. I think it’s like they are missing a “theory of mind” for how software works. We have developed this intuition by either using or creating software for a long time.
Like often people are scared to click random buttons or explore menus. We have used software for a long time, so we know that any destructive act would be proceeded by an “are you sure?” dialog box.
I think I would be in their shoes if I ever tried to change the oil on my car. I may be able to follow a procedure. But I would not deviate from it. It would not try to pop the covers off unrelated parts to peek into the alternator, for fear of electric shock.
Anyways, my rambling point is ideally we people would make themselves more familiar with how to understand the technology they work with everyday. But we should have empathy for people who have not contorted their minds to understand the peculiar behavior of software.
Because when they break things they lose hours of work they cannot afford to lose. My parents had a small business and if my mom mis-clicked somewhere and deleted some important thing, she would enter a panicked rage as she'd lose half the day with some remote tech support.
She would never touch a computer after work, because it just stressed her, she found them absolutely unbearable, also because the tech support was unbearably obnoxious.
> I think it’s like they are missing a “theory of mind” for how software works.
This is a delightful way to put it! I have older relatives that can't really tell apart where a website ends and where their operating system begins, due to not really knowing about the client server model (even when it's explained to them) and I worry about the generation that will also just grow up with phone apps.
That's why I actually liked the classes back in school where we had all sorts of office program tasks to do, though I wish some of the later stuff from university would also be more widespread. Probably not programming for everyone (unless something practical with like Python), but just a few lessons on how the Internet works.
> she was acting like I had pushed her onto debian or something.
I've actually upgraded elderly people's computers from various Windows versions to various Ubuntu versions. I always tell them it's the newest Windows version, and very rarely have seen any problems. No joke, one of the biggest issues I saw was finding a solitaire program similar enough to the windows solitaire program.
Not @ you in particular, but theres a rogue "computer guy" in my dads town who literally goes around "fixing" elderly peoples computers for a few hundred bucks, all he does is format them and install linux mint.
My dad, who is a little bit technical from a long time ago, spends a lot of time helping distraught users reinstall windows (to get some application going that the brains trust of the middle of nowhere australia cant get going on linux) or just helping people recover photos etc from cloud backups.
For what it's worth I've never been paid for that service. And I go to great trouble in enumerating the applications that people use and the files that they access, including cloud files.
That is seriously how some people navigate through life, not just computers.
Like if they want to go to a certain office in a building: My "instructions" would be to go to department foo, room A5.303 (made up number). Their instructions are something like "in that red brick building, take the elevator to 5th floor, turn left and then again left into the hallway, second door on the right side".
If they navigate UI the same way, then of course every element that is moved slightly to a new place or grouped differently will cause their mental model of how things work to collapse.
Your equivalence seems widely inaccurate to me. I too sometimes have a mental model which is "red building, third floor, left then left" because that’s spatial memory. A random room number tells you absolutely nothing about where you actually have to go. I can (and sometimes do) remember one and will use one if it’s all I have but that’s just random information totally different from actual direction.
I think your comment actually hits the heart of the issue. Most software developers I have met have this kind of disdain for their users familiarity and ability and act like they are above that. Most confuse their own learned familiarity when they were young and could tinker with some kind of superior ability. Well, just go read any post here about systemd, the syntax of Ocaml, or the new ux of any popular framework and you will see that we are all the same.
Accepting change takes time, especially if people had no desire for the change. Most people don’t care about their computers. They are for them a necessary impediment in the way of doing what they want to do. That’s why they dislike change: it looks like work for no upside. A major part of change management is explaining to people where the upsides are for them, helping them operating in the new environment and giving them time to process.
> Most software developers I have met have this kind of disdain for their users familiarity and ability and act like they are above that.
This is absolutely true.
For what it's worth, "so-and-so department, room 123" is only a helpful mental model for navigation if you have additional knowledge not present in that description: that the department is probably contained entirely in one specific building, where that building is, that the rooms are likely numbered sequentially following a pattern that also indicates the floor they're in, etc. By itself this is an address whereas "red building, third floor, left then left" is a turn-by-turn route.
In terms of UIs, the latter is more like "left-hand side of the menu bar, 'Edit' menu I think, half-way down the screen, something like 'Modify' or 'Change' or 'Transform' or something, then third item in the list, something like 'Skew' or 'Bend' or something" compared to the first just being "Transform -> Skew". The former is fuzzier and the exact labels don't really matter because you'll know them when you see them. The latter is extremely precise but tells you nothing about how to actually find what you're looking for (is it in the 'Edit' menu? is it in a right-click context menu? etc). It's only really useful by itself if you have a command palette.
> I think your comment actually hits the heart of the issue. Most software developers I have met have this kind of disdain for their users familiarity and ability and act like they are above that.
That is not what I meant. I don't say those users are worse for it, rather that user interfaces of course need to accommodate such approaches. Such as placing a notice at the old location which directs users to the new location (like when a brick-and-mortar-store moves).
Also when performing tasks routinely, of course I have muscle memory and will get slightly irritated when the newest function absorbed into systemd does things differently than the thing which it replaced. I have no desire for or against systemd, but I understand that this is how the world revolves and that one has get along with the times.
> Most confuse their own learned familiarity when they were young and could tinker with some kind of superior ability.
It is just willingness to learn. How you learn and how fast you learn is of course age dependent, but the attitude towards learning new things and breaking out of routine is what matters more.
It's also how IT professionals navigate software interfaces!
I'm amused by the amount of flak you are getting here.
Being fast and efficient with any piece of UI of course means to replace navigation and discovery with memory and intuition, at least in part!
Not just keyboard shortcuts, also the placement of UI elements, latencies, learning quirky behaviors... being fast with software means not having to discover features.
Similar to labeled controls on a physical device, just in a simulacrum.
>That is seriously how some people navigate through life, not just computers.
This "treat everything like a ludicrously complex black box that almost nobody can understand" for any given subject are the lowest common denominator that laws get written to.
This is maybe down to using the same cognitive pathways to navigate SW that we use to navigate the real world: way points en route to a destination.
Whereas those with more understanding of the system can deploy other methods that are a more direct fit to that structure.
The problem is, then, that if you change the terrain (e.g. new os) you also need to change the map/route in the heads of all those users without system knowledge.
Even some of those with system knowledge might STILL preferentially deploy their innate navigation cognition.
I also had an encounter with government software...
I think the long-lasting solution will be to move to a web-based application system, instead of depending on Desktop applications made for Windows or Linux. Using a web app system, the government only has to concern itself with proper development and maintenance of servers and web apps, and the public workers can use any operating system with a web browser.
Considering I'm working as a contractor for a local govt, working on a system that probably should be deployed to linux/containers, but I'm stuck on a locked down copy of windows with deployments going to Windows Servers. I can't even use WSL/Docker on my issued laptop.
The transparent proxy for all http(s) requests has been fun for some cli tools as well.
isn't TFA about how the problem is cloud-infrastructure? web-app will only make that worse unless you're thinking something like adobe air.
Also, as others already noted, somehow the web is uniquely terrible at maintaining compatibility across not only different platforms but different versions of the same platform and even different versions of the same browser, despite being built largely on top of high-level languages that all claim to enable effortless portability.
I agree with this. When it comes to domain-specific b2b software, pretty much everything that doesn't require native resources or performance inaccessible from the web platform should provide a web-based frontend (even if just as a minimum).
And that would be better why exactly? Bloated browsers trying to phone home, even if only to search for updates, or denying access because of overblocking DNS, false software security signatures, whatever? Ja. Klasse! I fucking know about Electron & Tauri. And V8, Node, Bun, TS. Hmmm so geil. Da steht mir der Schwanz steil. Oder auch einfach nur die Haare zu Berge.
I’ve come to find that change is not always good, and resistance to change can be good for those that resist.
For example, our German team knew the software that headquarters was going to deploy to them was broken, so they secretly continued to use their old existing process. When the new software was deployed “successfully”, the same software was then deployed in the U.S. and broke horribly. The Germans were told that their responsibility was their region (only), so they literally took care of themselves and refused to be forced to be the guinea pigs.
In addition, I’ve come to learn not to immediately upgrade my OS at work, if I can help it. Typically, being an early adopter means you can get some nasty bugs that sometimes can take days to figure out. No thanks.
Funny thing is, this was not a one-time occurrence at all.
Unfortunately, I didn't keep my personal photo around.
Here are two more... no hotlinks as I'm on mobile and web Reddit + iOS apparently have upped their annoyances to prevent copying image urls in Safari, and I won't bother more rn:
The French state is one thing, the Polytéchnique is another. My impression is the old-school network administrators at French universities are fiercely protective of their de facto right to make technical decisions regarding equipment and software. So this part surprises me.
> Having worked for the French state and wrestled a few times with its IT services, I can tell you that the reason for choosing Microsoft isn't cost, or "efficiency
While this is generally not wrong, the French state still uses and creates open source software extensively. Gendermerie Nationale have their own Linux distribution ; Ministry of Foreign Affairs runs Debian for diplomatic personnel. Multiple pieces of public software was developed out in the open (including the government SSO and the COVID tracking app).
So while there is definitely bad, it has been getting progressively better over the past ~10 years.
Switching from a working, integrated solution (SAS) to a untested collection of random scripts (R or Python??) is a horrendous idea, and I would be willing to bet that the people in charge of implementing this brilliant move saved the French state quite a bit of money by blocking it.
If they had started with the R or with Python, that would have likely been a better solution - but that doesn't mean that switching, especially just because some new employee convinced the director they'd save money with a half-baked idea, is not.
I work for a company whose entire business is basically replacing various dedicated internal tooling scripts with an integrated, extremely costly solution, and we mostly sell to other businesses. Most people who buy this are extremely happy to get rid of their kludge of scripts, and we very rarely feel pressure from such alternatives (our other B2B, also extremely expensive, competitors, are the main thing we get compared with).
Specialized software often works much better, and makes a lot of sense past some scale, than in-house solutions. Especially if there are no big geopolitical risks - which there aren't with SAS for the French government.
- The division doing the change wasn't doing anything critical, mostly yearly statistical analysis reports, and cost estimates for new laws. The move could have been done over a year or two. Most public workers aren't overwhelmed with work, to put it mildly.
- SAS language is horrendous, the IDE is stuck in the 2000´s, and everyone was doing SQL queries inside it as a result. SAS is obsolete compared to what R or Python can do. Career-wise, it's a dead end for workers who use it.
- The French state is heavily indebted, so any cost-saving measure is welcome to reduce the burden on the taxpayer.
But everyone experienced in such a "migration" would challenge your flawed impression that there would be cost saving. SAS is in fact used in many serious companies (I know it's used at the fraud department of Bouygues Telecom, for instance) so you're not gonna dead-end yourself all that much, you can do the python+r stuff on your own probably, as long as the reports are there, and call them from SAS if you cannot stand it.
Meanwhile, nothing has to move, nothing has to change, and your non-critical division will continue to cost exactly as budgeted instead of going through unknown waters just because maybe switching to something new might be ... cheap eventually ?
Plus this has nothing to do with Microsoft really, as most cost is in Windows and its integration. It's hard to replace the global policies of Windows by an alternative.
What are the costs of using python/R? I'm not talking about the full french gov, but a single division (around 25 workers), for which the procurement is done.
This is in reality a clear cost-saving measure, as you may have to pay one time for training, which is however done by another division of the French State (the statistical body, INSEE), and those career workers don't get bonuses or are paid overtime anyway.
Afaik other "big corps" are moving away from SAS as graduate only know Python/R nowadays, which are better and more flexible tools.
Improving tooling and reducing costs are just a thing for private companies. Yeah, the horse carriage drivers weren't happy when the gasoline engine appeared. Imagine the switching costs! There are still horse carriages around! Horses are reliable! You can drive a car on your own!
IDK man, getting some department out from under the thumb of the big org's IT is pretty much SOP when BigCo has a "oh crap we gotta compete in this niche, spin up an internal startup" moment.
Yeah, the plan is insane if you're talking about the entire french government.
> Especially if there are no big geopolitical risks - which there aren't with SAS for the French government.
Are you serious? Especially when officials working for well-respected international organizations are being sanctioned for their work out of political reasons? This entire article wouldn't exist if there were no geopolitical risk.
From the article:
> “Carniaux … was asked by the [French Senate] commission to guarantee that French citizens’ data hosted by Microsoft would never be transmitted to foreign authorities without the agreement of the French authorities. He replied: ‘No, I can’t guarantee that,'” the report stated.
If this isn't geopolitical risk, I don't know what is.
It's clear that your stated job constantly exposed you to all the talking points of mega cloud platforms. But there are negative consequences too, and they're very real.
Oops, I've made a huge mistake here, apologies. I confused SAS for SAP, the German business stats company. My post was entirely from this wrong perspective, that it would be replacing EU business software with open source software.
In light of the new understanding, I fully agree with you that it was extremely shortsighted and sad to have tech leadership oppose a project which directly gives a US company access to such sensitive data with open source alternatives.
Even when I see "so and so place is switching to X" I always think "uh, do your current Microsoft people ... know how to do that?"
I'm all for training and making the switch, but you gotta get your teams motivated / what they need to do the job. If not IT can resist like few other orgs.
Everything else aside moving to R and Python to escape the USA hegemony means you are still dependend on Python Software Foundation and R Fundation. First is American. Second is based in Vienna, Austria yet still has big American presence among its members. So you end up still dependending on USA. Same with most free and open source. Even if its authors may appear to be European, they may turn out to be Americans (Linus Torvalds) or work for US company anyway (Guido van Rossum)
This addiction to Microsoft is everywhere and it's terrible for everyone involved. So many small orgs and NGOs paying through the nose for what can be done for free with Google Docs & Sheets.
Using proprietary software that is given away for free is even worse than paying for proprietary software - at least you have a contract in the latter case.
Governments should pay software engineers and system administrators (or infrastructure engineers if you like) to build and run their systems, ideally open sourcing whatever makes sense
Yeah, but Google Docs and Sheets are really worse than Word/Excel for many uses/users. I’d be pretty peeved if my employer decided to save figurative pennies by making me use Google Docs/Sheets.
And Google Docs/Sheets isn’t really free once you need business functionality. Depends what exactly what your feature requirements are, but Google Workspace pricing has about the same range as Office 365 Enterprise. ($10-$50/user/month)
Sounds like they were honest about lying, less confusion than lying about lying and helps prevent the unwanted outcome of accidentally succeeding. Not succeeding helps with job security, the problem justifying your job will continue to exist. Since such skills are in demand at all levels of bureaucracy successfully failing to deliver at a lower level will open up opportunities to fail at a higher level.
Honestly, when management comes up with some silly idea like replacing a perfectly working piece of software that's core to your business with a completely different alternative that is not giving you anything but being slightly cheaper, caring about your job involves stopping this silly change from happening.
Caring about your job involves being informed enough to not open your organization up to abuse from the likes of Oracle and Microsoft. That would save your customers, or fellow citizens if you work for government.
While I still stand by my sentiment to a small extent (it's often people who do care that actually oppose management initiatives, and proprietary business software can be superior to OSS alternatives, in certain areas), I was working under the mistaken impression that this was about replacing SAP, a German business solution, not SAS, the American company.
I should say that I think the real solution here would be to create an EU wide business software public company (or even EU agency), which would build and maintain business software products for EU governments, preferably open source. There is so much commonality in these areas that it's absurd to not build a common core, and later provide consultancy and support services to member states for deep customization, from experts who understand this common core. This is basically the IBM, Oracle, SAS, SAP, or even Microsoft B2B business model, so not some pie-in-the-sky idea.
I do believe that having some unified organization that can provide support and expertise on how others have solved similar problems would be much better than having every municipality and institute in the EU develop its own bespoke OSS-based solution, or, worse, buy from a private company that can just extract a profit from the massive cost savings of deploying substantially similar solutions for hundreds of customers, instead of passing on the cost savings.
ok so "funny" but historically, it has been worse. Like the legend of a warrior King in a region towards modern Hungary.. the school teachers would be ordered to stop work when the King and his military advisors rode on the main road for inspection. All the teachers were required to stop and clean the litter by the road as the King passed by. Perhaps the implication is -- if you do not have my military protection, then your school would be burned by invaders? actually not wrong in some places, but can you blame that scene on "capitalism" ? Want to guess the expenses of the armor and servants for the King and their party? probably more coins than were spent on teachers, who were paid in farm produce and exchange services?
Depends on what you're doing, to be fair. R would probably be a better choice as it's more likely to have all the statistical stuff in SAS. Python might make sense if you want to integrate it into web/other systems.
It's definitely not an easy lift and shift though, as the OP made it out to be. Particularly when you're integrating with other departments it's vital to have perfect compatibility as otherwise you'll break the other departments workflow.
It's that they only know Microsoft, they don't want to learn something else, and if there's a problem, it's Microsoft's fault, no theirs, so they don't have to deal with their own incompetence.
If you want an anecdote, we were working with SAS, a statistical software which required costly licences (more than a million € for a few dozens of workers). I suggested to switch to R or Python to the top director, who agreed.
First meeting with the service in charge, the chief opens with "ok, we are asked to change, but the goal here is to show that we tried, and found that it's not possible."
I resigned a few months after, as everything was in the same vein.