Spinuzzi network-6&7

Editor's Notes

• #2: Hello, everyone, this is Clay Spinuzzi again. And we’re up to the last two chapters in Network: “Is Our Network Learning?” and “How Does Net Work Work?” But we’ll also be covering two heuristics that are not in the book, showing how you can relate different sorts of disruptions and get an overall picture of the systemic issues your users face. So in both senses – in terms of the book and in terms of our analysis – we’ll be pulling it all together.
• #3: So here’s the value you’ll get out of today’s session. Notice that these are grouped into two sets. The first set of points relate to additional material that’s not in the book. We’ll look at two additional heuristics that help us to better understand work. Operations tables help us to examine micro-level aspects of people’s work and especially the operational breakdowns that they encounter. Contradiction-Discoordination-Breakdown Tables allow us to relate all of the heuristics that we’ve developed so far, giving us a systemic understanding of the pervasive issues people face in the activity we’ve studied.Based on this work, we’ll pull the heuristics together into a recommendation report. I’ve supplied an outline and some sample recommendation reports to guide you as you write your own. The second set of points relate to the last two chapters of the book. We’ll examine how net work demands different learning techniques, and we’ll talk about how those techniques surfaced at Telecorp, using activity theory and actor-network theory to understand why these techniques were needed. Finally, we’ll wrap up the book by talking about implications for managers, workers, and researchers/consultants.
• #4: So let’s get started. First stop, our microlevel heuristic, operations tables. For more detailed instructions, see my book Topsight.
• #5: So let’s talk about operations. You may remember this discussion of levels from our very first slide deck, when we were talking about different aspects of genre. I mentioned that we could understand genre – and activity – at these three different levels. The macro level, the level of cyclical activity, functions at a time scale of years or decades and involves culture and history. We examined this level with two heuristics: activity system diagrams and activity network diagrams. And as we saw, the macro level involves a certain kind of disruption: systemic contradictions that build up among and within components of the activity.The meso level, the level of conscious goal-directed action, functions at a time scale of minutes or hours. It involves tool use. If you stop someone in the middle of their work and say “what are you doing?” they’ll typically respond at this level: “I’m writing a report.” “I’m updating my Facebook profile.” We examined this level with three heuristics: genre ecology models, communicative event models, and sociotechnical graphs. And this level involves another kind of disruption: discoordinations that build up between genres that are poorly related or that interfere with each other.The micro level, the level of unconscious operations, functions at a time scale of seconds. This level involves habits and unconscious skills that we pick up and use without thinking. And we’ll examine this level with just one heuristic – an operations table – that will let us detect and examine microlevel disruptions: breakdowns. Let’s discuss what this micro level involves first. Then we’ll get to some examples of breakdowns.
• #6: (picture: gearshift)The idea of operations comes to us from A.N. Leont’ev, one of Lev Vygotsky’s students. So this is an activity theory notion. Leont’ev explained that operations begin as conscious, tactical-level actions. For instance, suppose you’re learning how to drive a car with a manual transmission (a stick shift). At first, you have to consciously focus on pushing in the clutch, moving the stick shift to the next gear, then letting the clutch out again. You may even look at the clutch pedal or the stick shift.When I was just learning to drive, for instance, I would actually look at the little diagram at the top of the stick shift to figure out where each gear was. Every time. I would also have tremendous problems moving from a full stop when parked on an incline. It’s a difficult skill: I had to take my right foot off the brake pedal, put it on the accelerator pedal and push down, and simultaneously let out the clutch pedal with my left foot. And I had to make sure that I didn’t stall the car or roll backward into the car behind me. It was terrifying!But after a short while, shifting gears becomes habit. You don’t think of it as a long set of conscious steps – you think of it as “shifting gears,” and you may even stop noticing that you’re taking these steps. The steps have become operations, unconscious and habitual. And you’re free to focus on other things: instead of pushing in the clutch and moving the stick shift, you focus on driving. When I was 15 years old, I was conscious of every time I shifted gears; now, decades later, I will complete long trips without even being aware of shifting gears.In your consulting work, you probably won’t be examining how people shift gears. But you’ll probably see other examples of operations. http://www.flickr.com/photos/brilliantmichael/4172218074/
• #7: (keyboard)For instance, touch typing. Using a mouse. Jotting a note. Using gestures on a touchscreen. Answering a phone. Each of these examples involves complex steps that people must first learn consciously, as actions. But at some point they become habitual. A touch typist doesn’t think about pressing the letters C, A, T. She thinks: “Cat.” And she types the word “cat” without even looking at the keys.But operations don’t always work well. Sometimes we encounter breakdowns.Photo credit: Clay Spinuzzi
• #8: Breakdowns are micro level disruptions: points at which your learned, habitual operations don’t work as planned. For instance, let’s go back to the car example. Suppose that you rent a car, and the clutch is firmer than your own car’s – or maybe you’re used to driving a four-speed and the rental car is a five-speed. Worse, perhaps your clutch starts going out, so your car won’t stay in gear! So your habits don’t produce the usual results.When that happens, you have to refocus on that which is usually habitual. You become conscious of the individual steps, and you have a harder time refocusing on the higher-level tasks you’re trying to accomplish (like driving).http://www.flickr.com/photos/brilliantmichael/4172218074/
• #9: We often encounter breakdowns when using tools or technologies that are similar to, but not quite the same as, the tools and technologies with which we are familiar. For instance, When IBM released OS2/Warp in 1996, it included a Shredder icon. Like MacOS’s Trash Can and Windows’ Recycle Bin, OS2’s Shredder allowed users to delete files by dragging them on top of its icon.Unfortunately, the Shredder worked like an actual shredder: it didn’t save the files for later deletion, it deleted them immediately. Users who migrated from the other two platforms were consistently surprised by this behavior because they were not thinking about the icon in terms of the metaphor: they interpreted it as OS2’s version of the Trash Can, and they used the same habits – the same operations – as they had used for the Trash Can. These operations worked well until a user wanted to retrieve a deleted file – at which point the user encountered a rather significant breakdown.http://turbomilk.com/blog/cookbook/icon_design/10_mistakes_in_icon_design/
• #10: Here’s another example from Slide Deck 1. As you recall, I examined how people used PC-ALAS, a database of traffic accidents with a DOS-based interface. But by the time I studied PC-ALAS, these people had moved on to Microsoft Windows. They were unfamiliar with the DOS interface, which displayed interface elements as a grid of characters. The interface elements looked familiar – they looked a bit like Windows’ dialog boxes, buttons, fields, and menus – and people could even use a mouse to control a cursor on screen. But the two interfaces had small, significant differences. When people tried to use the operations they had learned in the Windows environment, those operations often led to breakdowns.In fact, I videotaped people as they performed their PC-ALAS tasks, then coded every interaction with the interface and every breakdown. And I spotted definite clusters of breakdowns around specific operations.For instance, in dialog boxes such as this one, I saw a cluster of breakdowns. Nine of the 12 participants encountered multiple breakdowns when they unsuccessfully attempted to enter input into dialog boxes. Their attempts failed primarily because PC-ALAS does not give the user precise control over mouse input: the mouse pointer is as large as a character, as are most of the input fields, so a successful attempt at input requires the user to click in exactly the right spot, with no leeway. In contrast, although Windows input fields tend to be smaller than PC-ALAS', Windows' mouse pointer is even smaller. Consequently, Windows users are able to be more precise in their input and at the same time did not have to make a "bull's eye" to activate an input field.Even more interesting, I was able to examine patterns of operations and breakdowns.
• #11: Here, the red triangles indicate breakdowns. The squares represent different operations related to different phases of their work. And as you can see, people varied in how frequently they encountered breakdowns and at what points in their work they encountered them. But the breakdowns were consistent enough that I could detect clusters around specific interface elements. Of course, coding the video data took forever. You can imagine automating this sort of data collection to some degree: for instance, by capturing how many times people hit the Cancel button, hit the Back button, or abandon and restart a sequence. Analytics provide a lot of potential material for understanding and mapping breakdowns.But you can also rely on observations and produce lower-granularity data about operations and breakdowns.
• #12: Here’s a simple heuristic that I have my students use with their observational and interview data. My students don’t videotape interactions, but they do observe people at work, interview them afterwards, and collect artifacts such as texts. And I tell them that as they do these things, they should be alert for breakdowns. Indicators might include:External indications of breakdowns: points at which the participant says “huh?” or “uh oh.”Points at which participants retrace steps.Points at which participants undo work – such as deleting a piece of text they have just written, hitting the Cancel button, using the eraser.Students then ask the participants what happened, and they examine any artifacts for clues. They also look at how participants typically recover from breakdowns. And they map these in operations tables like the one above. These operations tables provide a lower level of granularity, but they still allow students to identify clusters of breakdowns. And they help them determine whether the participants have established common ways to recover from these breakdowns.
• #13: I mentioned a moment ago that you might also be able to extract more detailed information on operations and breakdowns from examining analytics. Here, examining social media and social web applications opens up a new set of possibilities. You can find plenty of literature on analytics, but I’ll refer you to a case we discussed in Slide Deck 4. Eric Ries discussed how, in addition to conducting in-house usability tests, he would check the analytics to see how users interacted with IMVU. He used A/B testing – serving slightly different designs to users at random – to examine how the different designs affected factors such as registration, activation, retention, and referral (p.152). Google is famous for similar data-driven design. That sort of design is beyond the scope of our discussion – not to mention beyond my own expertise – but it opens up possibilities for examining breakdowns at an analytic level. Of course, analytics will not necessarily give you enough context to interpret operations and breakdowns. If you go the analytics route, I suggest pairing it with targeted observations and interviews so that you can properly triangulate the data.In any case, getting to the level of operations – and examining breakdowns – can give you a better idea of where users get stuck.http://www.imvu.com/
• #14: And this brings us to the last heuristic we’ll discuss. Throughout this slide series, we’ve been developing various heuristics and using them to detect various disruptions. Now it’s time to relate these disruptions.
• #15: Here are our heuristics along with the disruptions they help us to map: contradictions, discoordinations, and breakdowns. We’ve examined these separately. But these three levels really aren't separate. All those second-by-second operations at the micro scale make up the minutes and hours at the meso level and the longer cycles of work at the macro level. The disturbances at each level are connected. So we’ll pull in disruptions from all of our other heuristics at macro, meso, and micro levels:
• #16: At the macro level, contradictions in activity systems diagrams…
• #17: And in activity network diagrams.
• #18: At the meso level, discoordinations in the communicative event model…
• #19: …in the genre ecology model….
• #20: … and in the Sociotechnical Grapn.
• #21: And at the micro level, breakdowns in operations tables.And that’s why we need our final heuristic, the Contradiction-Discoordination-Breakdown (CDB) Table.
• #22: Think about it this way. Imagine you go to the doctor because you have certain symptoms:Severe and sudden sore throat.Pain or difficulty with swallowing.A high fever.Swollen lymph nodes in the neck.Swollen tonsils.If your doctor is incompetent, he might treat each symptom. For instance, he might tell you to swallow ice chips to soothe your throat and take aspirin for your fever. But if your doctor is competent, he will relate these symptoms and determine the systemic problem – the illness – that he should treat. If he does this, he will quickly determine that you have strep throat, and he will prescribe antibiotics.We want to be the competent doctor. In other words, we want to relate these disruptions and determine the systemic issues that cause them. If we don’t – if we simply treat the symptoms – the underlying issue may simply cause different symptoms elsewhere. http://www.flickr.com/photos/thirteenofclubs/5520512981/
• #23: Looking at the individual levels, we might find it hard to make connections. But pulling the different disturbances into a single table helps us to make these connections more easily, letting us name the different issues that our participants face. CDB tables represent findings in our research, findings that point the way to recommendations. For instance, in Slide Deck 1, we saw that students who were familiar with geographic information systems were totally flummoxed by GIS-ALAS. As we examine the contradictions, discoordinations, and breakdowns, we can see relations between them, and we might discover that they are all part of the same systemic issue: representations from two different activities simply don't mesh for the participants, and participants consequently have to spend a lot of time bridging between them. Of course, what makes this work tricky is that you may find multiple disruptions at each level, related to multiple systemic issues. For instance, you may find multiple contradictions, and they’re not all related to the same discoordinations. To untangle these, sometimes you’ll have to dive back into your source data to see if you can find connections among these. But – critically – you’ll need to be able to point to evidence for each connection you make.
• #24: For instance… (read)These all have different roles to play…
• #25: And the result might look something like this. Notice how each model plays its role; it gives you different parts of the story – different pieces of proof that help you to make your argument about the systemic issue.So what should we call this systemic issue? We could summarize the table in one sentence.
• #26: Here it is. The phrase summarizes the systemic issue, and then you can detail and justify it with your models and specific evidence. In fact, let’s make this even clearer:
• #27: Yes. This simple statement, derived from our CDB table, is our finding. In fact, if we write a recommendation report, we can turn the statement into a heading. And we can justify this statement by drawing on our CDB table and our other models, and further, on specific instances that served as basic evidence for those models.In fact, between now and the time we see each other in Oslo, I’ll have you read an example outline of a recommendation report as well as a few sample reports. You’ll use these as models for writing your own recommendation reports, based on your own work throughout this class. And I’ll give you detailed feedback.Our goal here is to use our theoretical tools – activity theory, actor-network theory, genre theory – together to generate an integrated understanding of the systemic issues involved at your site. We want to look past the symptoms to the disease. And that brings us to the systemic issues in Network Chapter 6.
• #28: What does learning look like in net work? My curiosity was piqued when I was studying Telecorp because I couldn’t initially understand how on earth the company held together. The constant turnover, the interrelated disciplines, the changing rules and units all seemed like a recipe for discohesiveness.And when I attended training, it confused me even further. As we’ll see in a moment, their training measures simply weren’t up to the job. Users used various learning techniques to learn different sorts of skills.
• #29: In fact,we’ve seen indications for a while now that people do use different techniques for solving problems. For instance, David Novick and his colleagues interviewed and observed office workers in 2007 and found that when these workers had trouble using software, they would turn to a variety of sources. Surprisingly, they tend to turn to software documentation last – even fewer people mention this option than mention giving up on the task! (Prof. Novick delivered these results at a conference devoted to software documentation.)Getting back to Telecorp, we find that they also use a variety of techniques for learning.
• #30: As we’ve seen throughout, but especially in the last slide deck, conditions at Telecorp changed constantly. In terms of operations: Departments split; departments adopted new rules and tools; regulations changed; companies were acquired; new services were introduced.In terms of people: Team members, as we saw in the last slide deck, circulated in, out, and through the organization (and the telecommunications industry in general). Turnover was high, people moved around different departments, people left.In terms of coordination, team members had to coordinate constantly: within departments, across departments, across companies, and with customers. That is, they had to cross departmental and organizational boundaries, communicating and coordinating with people outside their specialties. And due to the constant operational and personnel changes, this coordination work was always in flux.Let’s discuss this last point a little more, because it turns out to be very important.
• #31: I mentioned vertical and horizontal expertise.Vertical expertise is what we usually think about when we discuss expertise. A specialist uses his or her existing base of knowledge as a foundation for more complex and detailed knowledge. That is, the specialist learns more about her or his job. This is how school is set up. Children learn about addition, and that knowledge becomes the base for more complex operations such as multiplication and division. Undergraduates receive a degree, then go to graduate school to learn more about their specialty. It’s also the preferred mode of learning in a hierarchical organization. As you may remember, hierarchies limit connections across different parts of the hierarchy. Ideally, you work among others with the same specialty, and as you learn more about the specialty, you advance.Vertical expertise is absolutely important at Telecorp. For instance, a switch tech should learn more and more about switches on the job. If he’s not, that’s a huge problem.But at Telecorp, this hierarchy has been overlaid by a network in which workers can contact each other across (and outside) the organization. In fact, they must do so: they must contact, communicate with, and coordinate with people who have very different specialties, expectations, and backgrounds. For instance, the switch tech is going to have to be able to communicate with salespeople, because they’re in contact with the customers. In the process, he has to learn a little about what they know and vice versa. They have to develop horizontal expertise.
• #32: According to one switch tech, this wasn’t easy. Telecorp was full of technical jargon, and some parts of it traveled farther than others. But Ricardo, who had been working at Telecorp for a while, had begun to figure out how to cross boundaries. [read]So Telecorp’s workers had to learn both vertically and horizontally. How did that work out? I got a consistent story across the entire organization.
• #33: They said that their training was minimal – and in the same breath, they said that it HAD to be minimal, since “this is a different kind of industry.” Here are some of the metaphors they used. It’s not hard to figure out why this was the case. Work at Telecorp was highly contingent and its conditions changed very rapidly. Furthermore, turnover was very high and even the workers who stayed often shifted to different jobs in the company. The functional units in the company also tended to change rapidly, as did the units at the other companies with which Telecorp had to cooperate. All of these factors meant that knowledge had a short shelf life. Workers had to learn how to handle contingencies, and they did this primarily by experiencing each contingency.And yet, Telecorp learned. Workers used several techniques.
• #34: First, and most common, was apprenticeship. A new employee would come into the functional area and shadow an experienced worker. They would begin taking on responsibilities, then eventually be expected to do the job themselves. This method of training was mentioned in a majority of interviews, covering nearly all the functional areas. It was especially used in certain functional areas: the Internet Help Desk, LD Operations, the NOC, Network Coordination, and Sales.This type of training was contingency-based. That is, it was guided by whatever work transpired on that day. This sort of training is great for learning how to handle particular contingencies, but not good for teaching general principles. And that was problematic, because the point of one-on-one introductory training was to impart procedural knowledge and conceptual tools for understanding specialized work in small functional groups. Statements of general principle, such as “you never ever do a partial connection,” were embedded in encounters within the activity rather than framed broadly and in relation to each other.By the way, in retrospect, I think this training technique helped me out considerably in my research. I studied Telecorp by shadowing people in the organization – just like apprentices shadowed experienced workers. So my participants were comfortable with my presence. Lucky me.
• #35: Telecorp also had a few formal training sessions: organized, separate from work, involving an expert trainer. The most formal and common of these were the two-week training sessions that Abraham, the senior manager of Customer Service, provided to new workers. These training sessions were meant to introduce new workers to the industry. And I use the term “new workers” loosely here: since the training sessions were held only occasionally, they were often attended by people who had worked at Telecorp for three or four months.I attended an iteration of the course myself. In a conference room, workers from the NOC, Data Entry, and CLEC Provisioning sat two to a table, each with a thick training book in a three-ring binder. The training book consisted primarily of sample printouts without commentary or explanation. All chairs were oriented to the front of the room, where Abraham had a whiteboard and a computer with a projection unit. Although workers were encouraged to ask questions throughout the session, interaction mostly followed the pattern of inquiry, response, and evaluation.But Abraham was also full of stories. He had an endless supply of them: Stories about how workers had challenged BigTel’s charges on particular services, how Abraham had ordered his own personal phone service, how business owners had complained about their employees’ use of the auto-redial feature, how an auto parts store had paid more than they needed because they weren’t aware of a specific feature. Even when workers were asked to do exercises (on paper, simulating the computer screens they would be expected to encounter), their results were compared to the results Abraham put on the whiteboard and framed with stories about the situations in which they would be used. Such stories continually contextualized workers’ learning within Telecorp, emphasizing the local aspects of the knowledge.The stories emphasized contingencies: trainers, and often workers, offered stories of specific customers facing complex, unique circumstances and testimonials about how these were handled. One of the most commonly used phrases in training was “nine times out of ten, it’ll be …”
• #36: Corporate training was mentioned by twelve workers in nine areas. Workers who interacted frequently with technical counterparts in the industry – such as workers at Bill Verification, Translations, and Alarm Management – often attended training provided by other telecommunications providers (such as BigTel), equipment manufacturers, and software trainers. They often brought back thick manuals and three-ring binders full of notes from these sessions – along with newly learned terms, genres, and social languages.Workers generally reported learning a considerable amount from corporate training, including social languages that affirmed and distinguished their fields and trades, creating linkages with others outside Telecorp. But workers had difficulty in circulating this knowledge to those who did not attend the sessions. For instance, Gwendolyn in Human Resources reported that her manager had gone to training in Chicago and was “excited” about it, but nobody had had time to learn from her. So, ultimately, corporate training did not circulate well – though doubtlessly some of it was later filtered through apprenticeship.
• #37: Documentation was mentioned by nine workers in seven areas.Sometimes this documentation was self-generated : Workers would jot down notes, then consolidate them into a manual. For instance, Angelina was in charge of Billing, and had been for ten years. But she was about to leave. So in the two weeks before he left, she “made notes like crazy” and developed preprinted tables that she used to record and monitor her work in three-ring binders. Sometimes a new worker would have to replace someone who had already left, as Oscar in Alarm Management did, and would have to piece together the system of the previous worker by looking at documents and asking other workers in tangentially related functional groups. It was “hard to get ahold of exactly what this position did,” Oscar told me. You may recall that I got a similar story from Jean in Accounts Payable, who had to use existing spreadsheets to calculate Universal Service subsidies, even though she had no idea what they were.And some documentation existed, but wasn’t well publicized. Individuals might bring back manuals from corporate training outside Telecorp, for instance. But as you recall, these individuals didn’t have a good way to circulate their new expertise. Typically these training manuals would sit on their shelves – but occasionally they would be found by a colleague, who would then read them. One worker, Russel in Network Optimization, reported that he found such a manual, spent the weekend reading it, and came back a “completely different worker.”
• #38: Like corporate training, computer-based training tended to focus on industry-level rather than local concerns and fixed concepts rather than contingencies. Computer-based training was mentioned by four workers in three areas: the NOC, NW Design and Inventory, and CLEC Design and Inventory. Rusty in CLEC Design described this training program:“how each central office is connected … what a POTS line is, what a digital four-wire is, and a little bit of history of where the phone systems came from. … ground starts, loop starts and all that other stuff. Basically it's just a crash course … on phone systems and stuff.”But workers learned in at least two other ways. And these are actually much more interesting.
• #39: “Trial and error” was all about contingencies. “Trial and error,” or some variation, was mentioned by 15 workers in 11 areas. When workers used this phrase or an analog, they meant attempting to complete a task through self-directed exploration, then applying these lessons to future tasks.For instance, Nathaniel at the NOC reported that although he was apprenticed to a particular NOC worker, she was not willing to “sit there and give you advice.” So his training was “basically just jumping in there and being willing to get your hands dirty.” Many of the most memorable metaphors discussed earlier – “sink or swim,” “thrown to the wolves” – describe the sort of uncertainty and contingency with which workers had to deal as they learned through trial and error. Like apprenticeship, trial-and-error learning magnified contingencies.At the same time, trial and error resulted in some interesting and useful innovations. For instance, Kim in Bill Verification reported that workers helped each other with software: they learned tricks "by accident," then taught these to each other. Leann in Network Optimization found that she could import carrier rates from Excel (a spreadsheet) into Access (a database), then export the text into a text editor so she could assemble it into scripts, allowing her to avoid typing in the information; she conveyed this procedure to the worker she was training. Although it magnified contingencies, trial-and-error learning often did result in stabilized knowledge that could then be routinized.
• #40: Finally, stories were everywhere. We already saw how Abraham used stories in the formal training he led, but these were used everywhere else too. Maybe the best example is the story of how Rex the dog dies. We discussed this case in Slide Deck 2, but let’s take a look at how they informed the NOC about it.The NOC's assistant manager sternly told the story about Rex’s death to the entire NOC. “There was nothing about a dog on the ticket,” he said. “You must note that.” This is how the NOC promulgated the rule to ask about pets and locked gates: through stories delivered sternly by a manager to overworked trouble ticket specialists, half of whom were on the phone and typing four notes at once. Through stories circulated among those workers during short breaks as they waited for the next call. Such stories mediated and regulated the network. Like trial-and-error and apprenticeship learning, these stories emphasized contingencies and provided resources for dealing with them. But stories were not concrete or durable enough; they didn't circulate broadly enough to stabilize the network sufficiently. That is, there was no guarantee that stories would be passed on, told accurately, or interpreted consistently enough to induce rules.So: we’ve just talked about many learning techniques at Telecorp. Taken together, what do they mean? Think about them in terms of vertical and horizontal learning.
• #41: If we lay these different techniques out in terms of their learning contexts and whether they support vertical or horizontal learning, here’s what we find out. Vertical expertise was supported by a mix of formal techniques (Telecorp training sessions, documentation, computer-based training, corporate training) and informal techniques (trial and error, stories, apprenticeship). That is, when learning within their functional groups, trades, disciplines, and fields, Telecorp workers had a set of resources, some of which provided stable long-term frameworks, some of which were more oriented to contingency. But their horizontal expertise – the cross-boundary coordinative work that is so common and vital in net work – was supported almost wholly by informal, contingent ways of learning: workers learned through apprenticeships, stories, and trial-and-error how to perform the boundary crossing that they had to do each day across multiple activity contexts. And, constantly, the contingency involved in this boundary work was magnified even more than in the vertical learning, while stable interfaces and procedures were barely discussed outside the context of specific contingencies.That’s a big problem.
• #42: At the same time, we have to remember that somehow Telecorp still functioned. The assemblage has greater capabilities than its parts. People’s horizontal training tended to focus them on contingencies and tactical reactions rather than strategic, principle-based work, but they still tended to figure out their jobs by observing how others used their resources – especially their genres – and by trying out those resources themselves. Trainees plus genres plus other resources plus connections equals competence.http://www.flickr.com/photos/welcome2bo/4337922625/
• #43: So how do we apply these lessons to social media? This is actually a hard question – at least initially. That’s because social media aren’t like the work at Telecorp. At Telecorp, people were relying on these specialists to link with each other, learn complex routines, and get things right. In social media, the end users are much more open to different possibilities. People don’t get fired from social media. Social media companies want to foster engagement, not squelch it, and they do that in part by encouraging more engagement and innovation.
• #44: They even tolerate uses that might upset potential stakeholders. For instance, here’s the Google Plus page of Bank of America, claimed by a parodist. This activity is creative, but it’s hardly the same as the organized activity we saw at Telecorp. Yet if we look at more specific cases, we can see instances in which people use learning techniques in social media.https://plus.google.com/111797291613060149488/posts
• #45: Lee Sherlock’s study of World of Warcraft is a good example. You’ll recall from Slide Deck 4 that Sherlock examined how WoW players engaged in “grouping,” or complex team play for taking on game challenges. Sherlock pointed out that grouping is undersupported in the WoW interface itself. But players developed tools – and techniques, which they documented in a game wiki. In this case, the social medium involves specific short-term and long-term goals around which players organized. And once you have specific goals – and objectives, in the activity theory sense – it’s only a matter of time before people develop tools, rules, and divisions of labor to achieve them. So we can clearly see players engage in learning techniques, including trial-and-error leading to documentation. They can also share stories about their contingencies.Sherlock, L. (2009). Genre, Activity, and Collaborative Work and Play in World of Warcraft: Places and Problems of Open Systems in Online Gaming. Journal Of Business And Technical Communication, 23(3), 263-293.http://us.battle.net/wow/en/
• #46: Other social media also support specific activites. LinkedIn, for instance. In Slide Deck 1 and 5, I discussed how LinkedIn connects to other genres to support a fairly specific activity: hiring and being hired in the labor market. Like job application materials, a LinkedIn page is highly curated. Since the activity it supports has such a specific objective, LinkedIn tends to foster specific rules, tools, and divisions of labor. So specific, in fact, that …
• #47: … people are writing books about how to fine-tune one’s LinkedIn profile. A whole set of documentation has grown up around it. Not just books, but blog posts, forums, and other means of lateral connections among people. The clearer the objective of the social medium, the more likely you’ll find a cohesive, overarching set of learning practices that surround it.
• #48: Contrast that with more general social networks such as Facebook and Twitter. These more general social networks tend to support different sorts of interactions, but they don’t have a single overarching objective the way that World of Warcraft or LinkedIn do. So people apply different metaphors and thus different sources of genre knowledge to them. One group might think of Facebook as a sort of scrapbook; another, as a site for self-promotion; another as a place to circulate the jokes they used to circulate in chain emails. I haven’t conducted studies, but I suspect the learning techniques look very different at these sites. People have different, sometimes multiple objectives, rather than a single cohesive one, so learning is also much more localized.In examining learning techniques in social media, perhaps the best thing to do is to go back to the activity system diagram. What objective drives the activity in which the social medium is used?
• #49: Well. We have finished discussing most of the book and most of the slides. Let’s get to the final part – the implications. What do we do about net work?
• #50: Workers must constantly coordinate, negotiate, build trust and alliances, learn, and cross boundaries. Here’s four essential skills for workers in net work:Rhetoric. Net workers need to become strong rhetors. As I said in Slide Deck 1: Rhetoric, it’s not just for liars anymore. Net workers sorely need to understand how to make arguments, how to persuade, how to build trust and stable alliances, how to negotiate and bargain and horse-trade across boundaries. In net work, which is intricately and unpredictably connected, workers can find themselves doing this rhetorical work with nearly anyone.Time management. And because everyone is at the border, because black boxes are in short supply and of short duration, anyone can potentially lay claim to another’s time. Workers must be able to adopt or adapt ways to deal with work fragmentation, including genres and rules that allow them to create their own stable transformations, their own black boxes, for prioritizing, organizing, and achieving work. That might involve learning popular time management techniques or participating in online communities that face similar problems; they certainly will involve examining, evaluating, adapting, and adopting the local innovations that coworkers have developed.Project management. Similarly, when everyone is at the border, border-crossing is constant and collaboration across functional groups becomes more pervasive. Consequently, workers must take on more of the work that used to be done by managers: planning projects, developing strategic and tactical understandings of their projects, becoming aware of the other projects in which their collaborators are embroiled. They need to become aware of and manage the overlapping work activities that largely share the same tools but different rules, communities, and divisions of labor. Adaptability. Finally, workers must be ever more adaptable. Being on the border means having to learn horizontally as well as vertically, having to understand others’ work and social languages and genres, having to forage expertly for information. It also means learning how to assess sources and arguments, learning how to determine who to trust and when. Adaptabilitymeans being agile enough to splice new components into a relatively stable, woven system.So those are the skills workers need.
• #51: Managers may take away further implicationsBlack-boxing.We’ve talked about black boxes in general terms. But managers specifically need to encourage stabilizing regimes. Liaisons are workers or positions that develop to provide stable connections across groups. APIs, like the application program interfaces used in programming, consist of routines, protocols, and tools that allow simple interactions to generate complex effects. APIs in net work might include genres and other boundary objects. Aggregations are bottom-up characterizations of large sets of information, enabled by “applications that aggregate individual work practices in order to depict relations among the work of group members” (Hart-Davidson, Spinuzzi, & Zachry 2006). Managers could consider deploying this sort of infrastructure, which trades control over characterization for insight into emergent understandings of work.Strategic thinking. I advocated project management skills for workers. Managers should support these skills, providing topsight for workers, recognizing that without resources for strategic thinking, workers can become bogged down in a reactive tactical stance. More than ever, managers must provide a persuasive vision for each project and sufficient feedback for workers to see – and take ownership of – that project.Training. And that brings us to training. As we saw today, Telecorp’s workers received support for vertical learning through multiple channels, but support for horizontal learning was restricted to informal, contingency-oriented channels. Managers should find ways to support horizontal learning across boundaries, through formal as well as informal training and materials. They should particularly focus on supporting the sorts of skills that I mentioned above: rhetoric, time management, project management, and adaptability.
• #52: So what are implications for researchers – and for consultants? I suggest two.One, that we should be careful about bounding the case. Net work is characterized by unpredictable and intermittent connections among widely distributed points: workers could at any moment find themselves talking with people in the next cubicle or the next state, people who could be coworkers, competitors, customers, or even former coworkers, friends, or relatives. Net work is not neatly segmentable into contexts or activities. So we must instead follow the actors and texts, the contradictions, the disruptions, and especially the genres. The heuristics we’ve discussed should help with this.In addition, researchers might consider what net work – and in particular multiplicity – might mean for characterizing the data collected in a case study. We need to be diligent about setting up feedback loops with participants, not just to get their side of the story, but also to understand how they’re segmenting the world. Perhaps we look at the screen and see Facebook, but they look at the screen and see specific components tied to different activities. We’ve got to ask, frequently and intelligently.
• #53: And that brings us to implications for understanding social media.As I’ve argued throughout this series, social media aren’t just happening out there on their own. People use them to support existing activities (such as staying in touch with friends, finding a job); to carry on offline activities in a new online space (scrapbooking, playing games); and to learn and participate in new activities (such as spreading Internet memes). Sometimes they use the media for all of these simultaneously. We have to examine how social media are part of intersecting activities – both spliced and woven.But social media can also be understood at different levels of activity. People who use social media use them for different activities, clearly. But they also use them in different actions: in conjunction with different genres, in different sequences. And at the operational level, they learn shortcuts and habits. By looking at the different levels of activity in social media, we can gain further insights into how social media work.
• #54: Okay, so here are our takeaways. [read]