LAK12: Monday morning

This is the first full day of the LAK12 conference.

Shane Dawson welcomes everyone, and outlines practical arrangements. Apparently there is only one toilet for all 210 attenders so tickets will be issued.

Simon Buckingham Shum and Dragan Gasevic introduce the programme. Simon talks about what makes LAK special: friendly, sharing, multidisciplinary and bringing researchers and practitioners together. Dragan outlines the submissions – LAK11 had 17 contributions; LAK12 has 40. All papers were peer-reviewed by at least three reviewers. Simon talks over the demographics – about 75%/25% M/F. Nationality of authors – US 54, UK 24, CA 20, GER 15, SP 15. About 50% education; less than that about 40%; 30% policy/sensemaking. Academics plenty; business here but not plentiful; a handful of government/public sector. And one other!

There’s the SoLAR AGM on Wednesday 2.30pm, and the reception/technology demo tonight 5-7pm.

Koerner Library

Janet Giltrow, Associate Dean of Arts at UBC gives a welcome. Asks for help and input from the conference and learning analytics, particularly around curriculum change. Talks about how technology is changing her research area – new ways of making knowledge for researchers and genre participants. Talks about her research in to an online forum, exploring e.g. co-locations of utterances of interest. Thanks the organisers, welcomes us to Vancouver and says she expects we will need the conference umbrellas!

Caroline Haythornethwaite welcomes everyone, introduces Barry Wellman from University of Toronto, key expert in online social network analysis.

Barry Wellman: The Triple Revolution & Networked Individualism

The Triple Revolution & Networked Individualism: Implications for Learning Analytics.

He’s excited to be at the birth of a new discipline.

He has a new book out – he holds up a copy with uncorrected page proofs, but it’s out with Amazon by the end of May.

Has been working in this field for about 20 years. Shows a 1960s version of high tech – a desk with a keyboard. [Later: I think it was a Hollerith Tabulating Machine, so more like end of the C19th than 1960s.]


  1. The triple revolution
  2. Underlying structural changes
  3. Consequences of networked individualism for learning analytics, and wider

Acknowledges his influences – his wife Bev, Caroline Haythornethwaite, Jerry Bruner, and George Siemens.

Three revolutions, intertwined:

  1. Turn away from groups
  2. Internet – personalization (you as a person), distance effects still there but weakened
  3. Mobile-ization – being available, things being available

The West Coast disease – “everything’s already happened”.

People are networked; they are connected not in groups but in networks that move around. The big challenge of moving from atoms to bits, but there’s a bigger one in moving from groups (place-based, densely-knit) to networks (more far-flung, sparsely-knit, multiple, partial). Average network size 600-700 for North Americans. Dunbar’s Number? Forget it; the data don’t bear it out, people segue from one network to another, and have way more than 150 ties than you find in primates or villages.

Traditional small groups – all observe and interact with all. Were some people moving through, itinerants. Archetypical Tuscan Village, or Lower East Side NYC c1900. ‘Fun with Dick and Jane’ 1950s primer, nuclear family unit. Recurrent claims that things are always falling apart – Thomas Jefferson (1784), Ferdinand Tönnies (Gemeinschaft und Gesellschaft, 1887) to Robert Puttnam (Bowling Alone, 2000), Sherry Turkle (Alone Together 2010). We keep looking back. Very little evidence that this view of deterioration is true.

June 2006, American Soc Review, J Miller McPherson et al “Social isolation” – but problematic, and not replicated. Media moral panic. “Only someone from MIT can think of MIT as representative of anything other than MIT”.

GloCalization. Most travel is place to place without much sense of what’s in between. Despatialisation starting in the 1950s/60s – phones, airplanes, expressways.

People have about a dozen to a score of very close ties. Friends tend to know each other; workmates less so. People have about 600 ties. We are moving to networked, less face-to-face surveillance, a lot more sousveillance (e.g. Wikileaks) and coveillance, where we all check each other out on Facebook, Twitter etc. Online interactions linked to offline. Networks quite sparsely knit.

The individual is the unit, not the school, not the family; the person operates in their own networks. We’re strongly aware that learning takes place well outside classrooms. The world is not flat, it’s lumpy.

US vehicle use has increased; airline boardings are increasing dramatically and steadily, with only small blips for e.g. late 70s oil shock and 9/11. Fruit is globalised (e.g. of looking at labels in local store). Society based more on individual connectivity than on ethnic, religious and sexual memberships. Change in norms – ‘Guess Who’s Coming To Dinner’ in 1967, shock of having an interracial marriage, vs 2010 commercial dating ad. Interfaith marriage unremarkable too.

Personal computers have increased, and home broadband. Personal internet revolution – instant access, if you know where and how to look. Email contact as frequent with people 3000km away as 3km. Greater speed and volume – we log on as individuals.

Despite media and evidence-free pundits, much evidence shows a vibrant networked world. Many Pew Internet Studies. NetLab’s Connected Lives studies. World Interent Project surveys, StatsCan.

Modal shift: integrating travel with ICTs. People interact with the same universe online and offline; they are intertwined. Face-to-face still non-trivial at intercontinental range. Time spent at home down; kids of 18 months enjoying laptops and iPhones. Families operate as networks, less as groups. Guesses the most common email is “can you bring some milk home”.

Telus study – Canadians and Technology Survey 2009 – How has technology changed your relationship with your immediate family. 51% no impact. 35% made feel closer/more connected; 7% less close/connected. 7% don’t know.

Reconfiguration of private and public spaces – revealing personal information.

It’s not dystopian, people aren’t feeling alienated; connectivity isn’t fragmented. Online is continuation of personal relationships. You will make new friendships here, which will continue online. Insistence on the wrong concept of digital dualism, separation of online and offline relationships. It’s just not so in the studies. We spend time choosing media, they fit together into an ecology. Face to face matters because you get fuller bandwidth.

Conservation of media choice – if you email me, I tend to email you back, not use the phone.

There is a change in education too. People are creating like mad online. Wikipedia. Networked learner – multiplicity of technologies, tools and supports.

(Photo via @MikeAtkisson)

Are learners having conversations online? (Yes.) Is Turkle Right? (He says she isn’t.) Are there real conversations? Is reciprocity always tit for tit – help for x leads to help on x’, but not on y or z. Duncan Watts (Everything Is Obvious) move from Yahoo to Microsoft Research – was “everybody knows that” within an hour of it breaking.

People function as networked individuals. Social networks are larger; more internet leads to more in-person contact. More work at home part-time – not full-time as the telework vision had it. Networks sparsely-knit, loosely-bounded. Transportation is becoming somewhat fungible with information gathering. Up to 15 years ago, I used to drive to a library, now it’s a URL. Bounded groups to networked individualism – routinized stability to stable instability.

Book plug again – Networked: The New Social Operating System, Lee Rainie and Barry Wellman.


Caroline: Comment on how networked individual, student, where are they going to find their learning?

B: They may not be learning but they’re getting information. Speculating (don’t have data): Wikipedia. They can buy their papers. People are getting much information from many different sources. When I lecture, I have to be careful about my factoids – the smart kids are using their MacBook Airs and checking my facts out. Where information comes from is really powerful. As years in school goes up, percentage of information goes down.

Someone 1: How to apply this stuff to education, using it to learn. Should we teach this? C21st skills? How to teach this? Building community.

B: I haven’t, but Howard Rheingold has a new book out called Netsmart. It’s clear, readable prose. Was going to be Crap Detection 101, then realised Walmart would never shelve it with that title and they sell most books. Applied cynicism, or digital skills – finding, evaluating – is really important. Much attention paid to social pathologies, rather than the millions of cooperative engagement. I don’t know how to teach those skills, but people in this room do. I’m a sociologist.

Someone 2: Migrating among networks, have you seen compelling examples of that? Seems that we’re compartmentalised among our different networks – discussions with your wife vs profession.

B: We’re laying out a research agenda. Two competing thoughts. One, we all stay in intellectual silos. Hypothesis two, there’s a lot of spillover. I don’t know of any study that looks at that. There is one – hidden away in my book – that stuff on Amazon ‘people who bought this book bought that’ – it’s network data. Looked at people who bought key books in the 2008 election – three clusters without much overlap (silos): Republicans, Democrats, Obama-only (separate from Democrats too!). I don’t have hard data, do you?

Someone2: Look in the innovation community. Talking about political beliefs in that example. But in communities using a particular idea, innovation moves that in to new area of application.

B: Started (invested in?) a few companies that were vapour stock. Had good ideas. But thought Google+ would take off, but it hasn’t.

Someone 3: Thinking about how internet has grown over time, running out of IP numbers, where do you see this going with human networks?

B: I lost money on this. Anyone who does know is lying. Proliferation of mobile, for more than a phone to a bunch of apps. Some people think tablets will take over the world. There are more people who just want to consume information than people like me. I was involved with the internet before it began, in 1976.  Only realised this year that not everybody wants to be a sophisticated user, they just want to get what they need. On the other hand, people really want easy apps. I have friends who take photos of every dessert they eat and share them; low skill production of information. We’re going ease of use, and a lot of having it available with you. Whether tablets, ultrabooks, iPhone 5s, Windows 8, I’m not sure.

Someone 4: This is a pretty new field, new area of research. What should we focus on in terms of learning analytics? Setting research agendas at the moment. What should we do?

B: Here’s my list I rushed through . Payoff for local vs distant interactions; reciprocity tit for tit; how info passes through networks; how people navigate multiple networks. That’s what I want you to do. What did you come up with?

Someone 4: Differentiate Learning analytics from process analytics, concern about pedagogy; are there normative, ethics issues.

B: Sounds like old-line ed school stuff. Don’t forget the classroom, but go beyond that. See how everyday life integrates with what goes on in the classroom.

Someone 6: Workshop concerned privacy. Does sharing of this impede?

B: I don’t curse as much in presentations as I used to, I assume that you are videoing it. I don’t say bad things about U Toronto administrators. We are doing coveillance on each other, checking up a lot more. Simon Buckingham Shum invited me to this conference, and I went looking at his stuff. Secondly, we’re getting some useful information about what’s happening in the upper stories of power, e.g. Wikileaks. But when I go to RateMyProfessors, even though I get a high rating – I did it myself – you get students broadcasting about institutions about what’s good and bad, can game the ssytems. E.g. Chinese gold farmers doing Yelp reviews for money. I won’t deal with Google, they’re getting too much damn information about me. Major new triangle of growth – surveillance, records transferrable in health, education. We all have skeletons, I’m not happy with that going on.

Caroline thanks Barry.

Session 1A: Social Learning Analytics

Tobias Lay introduces the next session on ‘Social learning analytics’.

Dan Suthers and Kar-Hai Chu: Multi-mediated Community Structure in a socio-technical network

Dan is talking about work done with his student Kar-Hai Chu.

Learning is increasingly technologically embedded. Fundamental question about the synergies between the settings in which learning takes place. Demands analysis at multiple levels, connected.

Different theories about what learning is, and about how being in a social setting fosters learning. Can focus on the individual, or small groups, or larger collectives. Different epistemologies, the process: acquisition, intersubjective meaning-making (process, not outcome), participatory (change of position in a community through legitimate peripheral participation). (Suthers icCSCL 2006).

Tapped In – SRI’s network of education professionals. Longest-running educational online community. Got 8y of data, 20k educators/year; 40-60 volunteer-run activities per month. Geographical metaphor. Chats, discussions, wikis, file sharing. Schlager – it’s not a community, it’s a network; multiple forms of participation and mediational means of association.

Empirical matter: don’t assume network constitutes one community; don’t assume that external communities are replicated within the sociotechnical system. Approach: identify clusters, interpret, note media used.

It’s multi-mediated; Choice of techs for communication reflects and reaffirms relationship between interlocutors (Licoppe and Smoreda, 2005) – example of who you phone in the birth room, and who you email later. They analysed artifacts empirically as nodes, the same as people, and that tells you what kind of things they are (actants).

Two strategies for fine-grained analysis. First, interaction – looking at uptake; nodes represent events, and people are attributes. Second, associogram, take people as nodes, artifact as node, event is a line between them. Leads to interesting analysis – cohesive subgroup analysis, see if can interpret them as communities. Last year’s paper has details on transformation between the two.

Show actors tied closely to discussions, files, etc. (HICSS paper by his student explore this more). One actor creating most of the discussion posting, one doing file uploading, and they were interacting through them – closely coupled but in an asymmetric way.

Selected 2y of peak activity – data tidying to deal with mismatched timestamps, missing ids – much work. Associogram had 40.5k vertices – 20k actors, 12k discussion, 6k files, 3k chat rooms. 230k edges – weighted by number of times connected (events) – actor->artifact->actor. Associations between people are always mediated. 20m events!

Average path length is 4.4, actor-actor path length is about half. Largely due to a particular room; but when take that out, average path length was 6.02.

Looking for cohesive subgroup with identifiable common identity. “Community detection” (modularity partitioning) algoirithm (Blondel et al). Is NP-Complete, so have to approximate. Computed and visualised in Gephi. Looked at properties of high degree nodes. Uses display algorithms designed to deal with large graphs like this. Found six top cohesive subgroups.

One cluster (P3)  – found algorithmically – turned out to be people from the same organisation, and top two by degree were chatrooms. Also used discussions. That turned out to be a Midwestern school district using it for CoP mentoring of new teachers.

The two largest clusters were harder. P1 the high-degree nodes were rooms – the reception chatroom was huge. Many highest ranked rooms are own by Tapped in and function to welcome newcomers. 18% of actors only associated with this; overwhelmingly chat based. Not a community, but network for LPP .

P2, top were human actants – two very active volunteers in the system, who regularly facilitate After School Online events.

There were also a myriad of small groups. e.g. ESL course run by Australian sheep herder.

Purely structural graph theoretic computations identified cohesive subgroups that have interpretations as communities.  But do need an overlapping community algorithm.


Someone 1: Could you interpret each group you found?

Dan: It wasn’t easy. There are 170-odd of them. Many of them were fairly easy to interpret. Top two were messier, had people with no strong connection, as they were coming in. First one was the transcendent network around those activities.

George Siemens: With so much interaction in online spaces, your analysis was text centric. have you looked at impact from multimedia perspective?

Dan: That’s where the contingency graphs came from, students interacting through a whiteboard. That’s where uptake idea came from, for cross-media stuff. We worked on an extended example, to capture interaction through whiteboard and chat, a manual analysis. Automated whiteboard analysis, some stuff done 10y ago, but haven’t applied it yet.

Dragan Gasevic: Much of the impact of the content itself, have you considered use of content analysis?

Dan: Surprised got such good results without looking at content. Not doing LSA, that’s important too, would like to work in to this hierarchy.

Bieke Schreurs, Maarten de Laat: Network Awareness Tool

“Learning Analytics in the workplace, Detecting and Analyzing Informal Workplace Learning”, OUNL.

Bieke frames her presentation. Works at OUNL, dept funded by Dutch government to support professional development of teachers, and research on that. Special focus on how teachers learn informally in the workplace. Tacit knowledge, complex, hidden – unaware of learning, solving work-related problems. Really formal professional development activities not very effective, don’t solve the real problems. So focus on lifelong learning, informal learning, to support teachers. Our research is not online; in Netherlands, teachers not online. They do online, but interactions with teachers happen face-to-face. Interested in translating this to learning analytics field, use it as a plugin online.

Really do practice-based research, with the teachers. Tool is also for the teachers themselves, so they can reflect on it.

Network Awareness Tool – a social learning browser. User-generated tool, web 2.0 technologies.

Theoretical bases: Networked Learning. Social Network Theory (SNA). Social Capital.

Data collected – face-to-face interviews, ask what problems they encounter, what they want to learn from. Get info on learning topics. Get them to draw their ego networks on the topics, articulate the quality of them.

Tool is online here (in Dutch).

Web-based tool shows network visualisations. Saw school clusters; wanted to see connections above that level. Tool will be demoed this evening.

We collect this data: how can we collect it online? Whole series of mostly practical data-manipulation questions at multiple levels. Managers use different words to teachers talking about the same thing; not much of a community between them.

Teacher can browse their learning. Learning as a process of value connection. Interventions needed can be discussed.

Plans – working with UK OU on integrating with SocialLearn as a plugin, with Simon Buckingham Shum.


Someone 1: What tool did you use for the SNA graphs?

Someone 2 (?Maarten de Laat): I did it myself in Javascript with JIT.

Someone 3: 1. Do you have examples of teacher or school use of the feedback? 2. How do you deal with the social sensitivity of this information? Do teachers feel comfortable with telling you e.g. who they don’t interact with? Tension between privacy and having information useful and therefore public.

Bieke: Teachers have personal login file, they create their own personal ego networks. We explain to them it’s a reflection tool, not evaluation – only for themselves. Until now, they don’t have problems with it. In one school, big reorganisation where people were losing their jobs, still didn’t mind filling in information for the research. Only use it in the school, in the classroom. We have agreements to use it only as they wish. People who are really personal learner, reflect on their own activities. Worked with teachers on visualisation, shapes – e.g. ‘I don’t want to be a triangle, or a square’ – didn’t encounters problems yet. Practical examples of the feedback: are using the tool in three projects, do feedback personally or in group, depending on project’s purpose. It’s about the topics. Example of teachers on classical topic, could see they were interacting in-group, saw needed to look out. We don’t provide feedback, only show the images and ask them what they see.

Leyla Zhuhadar and Rong Yang: Cyberlearners and Learning Resources

Rong Yang is the speaker, teacher in Computer Science Western Kentucky University.

One main purpose of learning analytics is to provide insight into process by which learners master their subject. We can provide helpful information and optimise the learning experience. Cyberlearners now have access to a great deal of learning resources. Causes difficulties for learners to get hands on the information they need, or appropriate it effectively. Open source platforms create opportunity for studying interaction between learners and the system, to dynamically improve their experience.

In the US – several open source platforms – MIT OCW; Harvard Library Open Collections Program; Stanford Introduction to Artificial Intelligence and Machine Learning. HyperManyMedia (HMM) distance learning platform at Western Kentucky, about 6-7y old.

Main RQs: How can we detect community of similar cyberlearners based on structure of huge social network. How can we present it visually to analyse behaviour? How can we use this to build recommendation system?

Mined HMM logfile. 750k users. Resources arranged hierarchially – colleges, courses, resources, individual media objects (10k+).

Majority of learners prefer visual interface (?vs search).

Can our cognitive system deal with such a vast amount of resources? Are cyberlearners able to find what they really want?

Limits of cognition – George Miller 7 +/- 2 limit on pieces of information to be held simultaneously in short-term memory.

We do give a semantic recommender system, but may not be enough. Can we help cyberlearners remember by linking or relating them conceptually to other resources?

Finding communities, using logfile turned in to graph. Vertices as users and resources, edges connecting users to resources. Used network visualisation to detect community candidates.

Example of social network structure of a weblog – 8k nodes, 14k edges. Wanted to simplify. About 10 communities of learners based on similarity (commonalities). A client (learner) belongs to one. Instead of one learner among 8k learners, are one among 800. Still large. Don’t want recommendation from all 800 people. Each learner has unique distance (weights) from resource; a dot ahead is a learner who is a bit more experience; a dot behind is one less experience.

Applied chunking hypothesis (Herbert Simon) – so each client is recommended with small numbers of users they’ve not interacted with before but ?are close.

Plan to implement this recommender system, evaluate.


Someone: How are those three nearest neighbours identified?

Rong: It’s a triad, three who are connected.

Dan Suthers: Are you empirically checking out the 7 +/- 2 thing? I think people can deal with more because we rely on external artifacts to extend our capabilities. Be careful!

Rong: We were just trying to limit the amount of information we recommend to our client. That’s just some kind of theoretical base.

Dan: Have seen this overused. We can deal with more than 7 if externalised in the right way.

Someone 2: Suggestion – open source is associated with Richard Stallman’s Free Software movement. Are those open source programs related to that? Open access is more appropriate?

Rong: Means it’s free of charge, learners can go in and use it.

Someone 3: Open Educational Resources! [is what they really mean]

Nazim Rahman and Jon Dron: Challenges and opportunities for learning analytics when formal teaching meets social spaces

Jon Dron says that the title is what he was supposed to talk about this. Nazim left just after writing this, taking software with him.

[Fun slides based on what look like Draw Something sketches.]

W Brian Arthur definition of technologies – the orchestration of phenomena to some use. Could be mental, religious – doesn’t have to be toys.

Hard/soft technologies. Hard – is embedded – iPhone, car, integrate a lot of components and pieces; also includes legal systems, processes in organisations. Stuff we don’t have to think about it. Soft – is done by people. E.g. paintbrush. Flexible, creative, adaptive. Want balance, just right between the two. Legal systems can be very hard; email is very soft, can use for lots of things, we are part of that technology.

Soft technologies can be used skillfully ; hard technologies may be used correctly.

All technologies are assemblages; they exist in relationship to, are made of, other technologies. We change them by adding to them. Kuhnian paradigms – we don’t get paradigm shifts, we stick things together.

Athabasca Landing – they are designing it at Athabasca. It has a social network but isn’t one. When we harden things, we leave spaces between. Many things done by email, or by massively over-process-ising a thing. The system is intentionally a very soft system assembled from hard pieces. For everyone – not just students, but researchers, administrators – anyone within a learning community. We hope learning will happen. Caters for what Terry and I talk about as three social forms – groups, sets and nets. Groups trad, nets known. Sets are not personal connection, but about things we’re interested in – interest communities, sites – e.g. Twitter, Pinterest – not about the networks. Relational databases we’re using are the archetype of sets. Many ways to learn, many other uses. It’s complicated, it all gets mixed up. Course tags, use following mechanism. Many variants, very hard to identify.

What’s going on? Is it working? Have 3k users today, some students, administrators, cleaners. Could be doing things informally. No unifying view. Much we can do. Need analytics system that can cope with the formal structure and the informal ad-hoc groups, sets and nets.

LASSIE – actually called AU Analytics. Freeform analytics. Not too smart on the site. Standalone version could be interesting, but Nazim has it. Filters people in groups, activities, what they’re commenting on – or/and their relationships in nets, and sets (tags).

Biggest problem is asking the right questions.

Some findings – people tend to use a subset of tools, and tend to continue that way. Students on a formal, hard course have a tendency to continue the tools used in that course. Matthew Principle. Can see what happens when you take people out of the network. Hub removal. We can disable users, but deleting them removes all their content, which is problematic. When you take them out, the network turns out to be scale-free; can take most nodes out, but some highly-connected ones out it destroys the network.


Simon Buckingham Shum: Is there any role for theory in designing social learning analytics?

Jon: Yes. What kinds of theory?

Simon: Examples to give distinction from SNA in an organisation – how shapes your thinking?

Jon: Excellent question. We bring whatever theories we have to the table, then have means of exploring them, though not testing. Looking at pedagogies in groups, socio-technical construction. In my courses, I say students, do what you like but remember to tag it. Based on my beliefs about ownership and self-direction. Others use a different set of things. Aiming for something that’s so soft it’s open to everything. Further question about how to design technology for learning: as an assemblage, include process, pedagogy, models. See as strongly interconnected. Build them from small pieces.

Simon: Assemblage has come up. Emergent design, community looking at this. Quite appealing.

Jon: Inspiration from variety of communities. Kevin Kelly, Brian Arthur. Complexity Theory, Stewart Cowthorne (??). Actor-Network Theory. Same kinds of problems from different perspective. Systems theories. We are among the few bringing this together as learning technology design.

Dragan: Soft technologies are great, but can be very frustrating. Allow creativity, support tinkerers. Frustrating for those who don’t like that. How do we design, manage the complexity, to address all the types?

Jon: The trick is making soft things out of hard things. Assemblage is core. No one way will suit everybody – or same person in different context. Price of flexibility is that soft is hard, hard is easy. Build them out of hard bits. Don’t require huge amount of thought. Technology that’s hard for one person may not be for the next; misunderstanding. A LMS is very soft to someone building it; quite soft to teacher; very hard to student. Understand different contexts. Key thing is make the pieces fairly large, hard, easy; put them together to make them as hard or soft are we think they need to be. Which is a separate question!

[My question – not time to ask- How are you planning to scaffold people to appropriate soft technologies when what you can do with it is, by its nature, not obvious?]

This work by Doug Clow is copyright but licenced under a Creative Commons BY Licence.
No further permission needed to reuse or remix (with attribution), but it’s nice to be notified if you do use it.

Author: dougclow

Data scientist, tutxor, project leader, researcher, analyst, teacher, developer, educational technologist, online learning expert, and manager. I particularly enjoy rapidly appraising new-to-me contexts, and mediating between highly technical specialisms and others, from ordinary users to senior management. After 20 years at the OU as an academic, I am now a self-employed consultant, building on my skills and experience in working with people, technology, data science, and artificial intelligence, in a wide range of contexts and industries.

3 thoughts on “LAK12: Monday morning”

  1. I was sitting at your table this morning during the keynote. You’re doing a great job summarizing and I’ve forwarded a link to your blog on to my colleagues who aren’t here as part of my summary. Thanks!

Comments are closed.

%d bloggers like this: