Microsoft unilaterally and quietly changed the spam filtering rules for Athabasca University’s O365 email system on Thursday afternoon last week. On Friday morning, among the usual 450 or so spams in my spam folder (up from around 70 per day in the old Zimbra system) were over 50 legitimate emails, including one to warn me that this was happening, claiming that our IT Services department could do nothing about it because it’s a vendor problem. Amongst junked emails were all those sent to the allstaff alias (including announcements about our new president), student work submissions, and many personal messages from students, colleagues, and research collaborators.
The misclassified emails continue to arrive, 5 days on. I have now switched off Microsoft’s spam filter and switched to my own, and I have risked opening emails I would never normally glance at, but I have probably missed a few legitimate emails. This is perhaps the worst so far in a long line of ‘quirks’ in our new O365 system, including persistently recurring issues of messages being bounced for a large number of accounts, and it is not the first caused by filtering systems: many were affected by what seems to be a similar failure in the Clutter filter in May.
I assume that, on average, most other staff at AU have, like me, lost about half an hour per day so far to this one problem. We have around 1350 employees, so that’s around 675 hours – 130 working days – being lost every day it continues. This is not counting the inevitable security breaches, support calls, proactive attempts at problem solving, and so on, nor the time for recovery should it ever be fixed, nor the lost trust, lost motivation, the anger, the conversations about it, the people that will give up on it and redirect emails to other places (in breach of regulations and at great risk to privacy and security, but when it’s a question of being able to work vs not being able to work, no one could be blamed for that). The hours I have spent writing this might be added to that list, but this happens to relate very closely indeed to my research interests (a great case study and catalyst for refining my thoughts on this), so might be seen as a positive side-effect and, anyway, the vast majority of that time was ‘my own’: faculty very rarely work normal 7-hour days.
Every single lost minute per person every day equates to the time of around 3 FTEs when you have 1350 employees. When O365 is running normally it costs me around five extra minutes per day, when compared with its predecessor, an ancient Zimbra system. I am a geek that has gone out of his way to eliminate many of the ill effects: others may suffer more. It’s mostly little stuff: an extra 10-20 seconds to load the email list, an extra 2-3 seconds to send each email, a second or two longer to load them, an extra minute or two to check the unreliable and over-spammed spam folder, etc. But we do such things many times a day. That’s not including the time to recover from interruptions to our work, the time to learn to use it, the support requests, the support infrastructure, etc, etc.
To be fair, whether such time is truly ‘lost’ depends on the task. Those ‘lost’ seconds may be time to reflect or think of other things. The time is truly lost if we have to put effort into it (e.g. checking spam mail) or if it is filled with annoyance at the slow speed of the machine, but may sometimes simply be used in ways we would not otherwise use it. I suspect that flittering attention while we wait for software to do its thing creates habits of mind that are both good and bad. We are likely more distracted, find it harder to concentrate for long periods, but we probably also develop different ways of connecting things and different ways of pacing our thinking. It certainly changes us, and more research is needed on how it affects us. Either way, time spent sorting legitimate emails from spam is, at least by most measures of productivity, truly time lost, and we have lost a lot of it.
Feeding the vampires
It goes without saying that, had we been in control of our own email system, none of this would have happened. I have repeatedly warned that putting one of the most central systems of our university into the hands of an external supplier, especially one with a decades-long history of poor software, broken or proprietary standards, weak security, inadequate privacy policies, vicious antagonism to competitors, and a predatory attitude to its users, is a really stupid idea. Microsoft’s goal is profit, not user satisfaction: sometimes the two needs coincide, often they do not. Breakages like this are just a small part of the problem. The worst effects are going to be on our capacity to innovate and adapt, though our productivity, engagement and workload will all suffer before the real systemic failures emerge. Microsoft had to try hard to sell it to us, but does not have to try hard to keep us using it, because we are now well and truly locked in on all sides by proprietary, standards-free tools that we cannot control, cannot replace, cannot properly understand, that change under our feet without warning, that will inevitably insinuate themselves into our working lives. And it’s not just email and calendars (that can use only slightly broken standards) but completely opaque standards-free proprietary tools like OneDrive, OneNote and Yammer. Now we have lost standards-compliance and locked ourselves in, we have made it unbelievably difficult to ever change our minds, no matter how awful things get. And they will get more awful, and the costs will escalate. This makes me angry. I love my university and am furious when I see it being destroyed by avoidable idiocy.
O365 is only one system among many similar tools that have been foisted upon us in the last couple of years, most of which are even more awful, if marginally less critical to our survival. They have replaced old, well-tailored, mostly open tools that used to just work: not brilliantly, seldom prettily, but they did the job fast and efficiently so that we didn’t have to. Our new systems make us do the work for them. This is the polar opposite of why we use IT systems in the first place, and it all equates to truly lost time, lost motivation, lost creativity, lost opportunity.
From leave reporting to reclaiming expenses to handling research contracts to managing emails, let’s be very conservative indeed and say that these new baseline systems just cost us an average of an extra 30 minutes per working day per person on top of what we had before (for me, it is more like an hour, for others, more). If the average salary of an AU employee is $70,000/year that’s $5,400,000 per year in lost productivity. It’s much worse than that, though, because the work that we are forced to do as a result is soul-destroying, prescriptive labour, fitting into a dominative system as a cog into a machine. I feel deeply demotivated by this, and that infects all the rest of my work. I sense similar growing disempowerment and frustration amongst most of my colleagues.
And it’s not just about the lost time of individuals. Almost always, other people in the system have to play a role that they did not play before (this is about management information systems, not just the digital tools), and there are often many iterations of double-checking and returned forms, because people tend to be very poor cogs indeed. For instance, the average time it takes for me to get recompense for expenses is now over 6 months, up from 2-4 weeks before. The time it takes to simply enter a claim alone is up from a few minutes to a few hours, often spread over months, and several other people’s time is also taken up by this process. Likewise, leave reporting is up from 2 minutes to at least 20 minutes, usually more, involving a combination of manual emails, tortuous per-hour entry, the ability to ask for and report leave on public holidays and weekends, and a host of other evils. As a supervisor, it is another world of pain: I have lost many hours to this, compounding the ‘mistakes’ of others with my own (when teaching computing, one of the things I often emphasize is that there is no such thing as user error: while they can make mistakes and do weird stuff we never envisaged, it is our failure to design things right that is the problem). This is not to mention the hours spent learning the new systems, or the effects on productivity, not just in time and motivation, but in preventing us from doing what we are supposed to do at all. I am doing less research, not just because my time is taken with soul-destroying cog-work, but because it is seldom worth the hassle of claiming, or trying to manage projects using badly designed tools that fit better – though not well – in a factory. Worse, it becomes part of the culture, infecting other processes like ethics reviews, student-tutor interactions, and research & development. In an age when most of the world has shaken off the appalling, inhuman, and empirically wrong ideas of Taylorism, we are becoming more and more Taylorist. As McLuhan said, we shape our tools and our tools shape us.
To add injury to insult, these awful things actually cost money to buy and to run – often a lot more money than they were planned to cost, making a lot less savings or even losses, even in the IT Services department where they are justified because they are supposed to be cutting costs. For instance, O365 cost nearly three times initial estimates on which decisions were based, and it appears that it has not reduced the workload for those having to support it, nor the network traffic going in and out of the university (in fact it may be much worse), all the while costing us far more per year to access than the reliable and fully-featured elderly open source product it replaced. It also breaks a lot more. It is hard to see what we have gained here, though it is easy to see many losses.
The one justification for this suicidal stupidity is that our technological debt – the time taken to maintain, extend, and manage old systems – is unsustainable. So, if we just buy baseline tools without customization, especially if we outsource the entire management role to someone else, we save money because we don’t have to do that any more.
This is – with more than due respect – utter bullshit.
Yes, there is a huge investment involved over years whenever we build tools to do our jobs and, yes, if we do not put enough resources into maintaining them then we will crawl to a halt because we are doing nothing but maintenance. Yes, combinatorial complexity and path dependencies mean that the maintenance burden will always continue to rise over time, at a greater-than-linear rate. The more you create, the more you have to maintain, and connections between what we create adds to the complexity. That’s the price of having tools that work. That’s how systems work. Get over it. That’s how all technology evolves, including bureaucratic systems. Increasing complexity is inevitable and relentless in all technological systems, not withstanding the occasional paradigm shift that kind-of starts the ball rolling again. Anyone that had stuck around in an organization long enough to see the long-term effects of their interventions would know this.
These new baseline systems are in no way different, save for one: rather than putting the work into making the machines work for us, we instead have to evolve, maintain and manage processes in which we do the work of machines. The complexity therefore impacts on every single human being that is having to enact the machine, not just developers. This is crazy. Exactly the same work has to be done, with exactly the same degree of precision as that of the machines (actually more, because we have to add procedures to deal with the errors that software is less likely to make). It’s just that now it is done by slow, unreliable, fallible, amotivated human beings. For creative or problem-solving work, it would be a good thing to take tasks away from machines that humans should be doing. For mechanistic, process-driven work where human error means it breaks, it is either great madness, great stupidity, or great evil. There are no other options. At a time when our very survival is under threat, I cannot adequately express my deep horror that this is happening.
I suspect that the problem is in a large part due to short-sighted local thinking, which is a commonplace failure in hierarchical systems, and that gets worse the deeper and more divisive the hierarchies go. We only see our own problems without understanding or caring about where we sit in the broader system. Our IT directors believe that their job is to save money in ITS (the department dealing with IT), rather than to save money for the university. But, not only are they outsourcing our complex IT functions to cloud-based companies (a terrible idea for aforementioned reasons), they are outsourcing the work of information technologies to the rest of the university. The hierarchies mean a) that directors seldom get to see or hear of the trouble it causes, b) they mix mainly with others at or near their hierarchical level who do not see it either, and c) that they tend to see problems in caricature, not as detailed pictures of actual practices. As the hierarchies deepen and separate, those within a branch communicate less with others in parallel branches or those more than a layer above or below. Messages between layers are, by design, distorted and filtered. The more layers, the greater the distortion. People take further actions based on local knowledge, and their actions affect the whole tree. Hierarchies are particularly awful when coupled with creative work of the sort we do at Athabasca or fields where change is frequent and necessary. They used to work OK for factories that did not vary their output much and where everything was measurable though, in modern factories, that is rarely true any more. For a university, especially one that is online and that thus lacks many of the short circuits found in physical institutions, deepening hierarchies are a recipe for disaster. I suppose that it goes without saying that Athabasca University has, over the past few years, seen a huge deepening in those hierarchies.
Our university is in serious financial trouble that it would not be in were it not for these systems. Even if we had kept what we had, without upgrading, we would already be many millions of dollars better off, countless thousands of hours would not have been wasted, we would be far more motivated, we would be far more creative, and we would still have some brilliant people that we have lost as a direct result of this process. All of this would be of great benefit to our students and we would be moving forwards, not backwards. We have lost vital capacity to innovate, lost vital time to care about what we are supposed to be doing rather than working out how the machine works. The concept of a university as a machine is not a great one, though there are many technological elements and processes that are needed to make it run. I prefer to think of it like an ecosystem or an organism. As an online university, our ecosystem/body is composed of people and machines (tools, processes, methods, structures, rules, etc). The machinery is just there to support and sustain the people, so they can operate as a learning community and perform their roles in educating, researching and community engagement. The more that we have to be the machines, the less efficiently the machinery will run, and the less human we can all be. It’s brutal, ugly, and self-destructive.
When will we learn that the biggest costs of IT are to its end users, not to IT Services? We customized and created the tools that we have replaced for extremely good reasons: to make our university and its systems run better, faster, more efficiently, more effectively. Our ever-growing number of new off-the-shelf and outsourced systems, that take more of our time, intellectual and emotional effort, have wasted and continue to waste countless millions of dollars, not to mention huge costs in lost motivation and ill will, not to mention in loss of creativity and caring. In the process we have lost control of our tools, lost the expertise to run them, lost the capability to innovate in the one field in which we, as an online institution, must and should have most expertise. This is killing us. Technological debt is not voided by replacing custom parts with generic pieces. It is transferred at a usurious rate of interest to those that must replace the lost functionality with human labour.
It won’t be easy to reverse this suicidal course, and I would not enjoy being the one tasked with doing so. Those who were involved in implementing these changes might find it hard to believe, because it has taken years and a great deal of pain to do so (and it is far from over yet – the madness continues), but breaking the system was hundreds of times easier than it will be to fix it. The first problem is that the proprietary junk that has been foisted upon us, especially when hosted in the cloud, is a one-way valve for our data, so it will be fiendishly hard to get it back again. Some of it will be in formats that cannot be recovered without some data loss. New ways of working that rely on new tools will have insinuated themselves, and will have to be reversed. There will be plentiful down-time, with all the associated costs. But it’s not just about data. From a systems perspective this is a Humpty Dumpty problem. When you break a complex system, from a body to an ecosystem, it is almost impossible to ever restore it to the way it was. There are countless system dependencies and path dependencies, which mean that you cannot simply start replacing pieces and assume that it will all work. The order matters. Lost knowledge cannot be regained – we will need new knowledge. If we do manage to survive this vandalism to our environment, we will have to build afresh, to create a new system, not restore the old. This is going to cost a lot. Which is, of course, exactly as Microsoft and all the other proprietary vendors of our broken tools count upon. They carefully balance the cost of leaving them against what they charge. That’s how it works. But we must break free of them because this is deeply, profoundly, and inevitably unsustainable.