The C3 meeting was held at UEF from September 12-15th 2015. It had four ingredients:

- a review of the current project status, identifying key issues for the final year of the project

- practical work and planning aimed at addressing topical concerns

- seminar activities in which we heard about other related projects and explored links between their agendas and 'making construals'

- reflection on the matters arising from the meeting

The practical work was carried out in parallel with the third Transnational Project Meeting which was led by Mike Joy and the other members of the project management committee representing all six partners.

The review focused on identifying resources that might be deployed in Virtual Workshops and how these might be incorporated in an open online course. Thematically interesting topics that had been explored in year 2 included: maths education (with particular attention to "mathematical resilience"), physical computing (in relation to experimental work on using the MCE with Arduino by Toivonen and Pope), resources relating to computing at school(including construals developed in association with the 'Computer support for computing unplugged' presentation at the UK CAS National conference in June 2016) and the potential role of making construals in software development (as illustrated in the way in which Nick Pope had re-engineered the MCE to exploit the observation-dependency-agency principles). Several initiatives and publications, both in process and forthcoming, that were developed at the end of year 2 served as background resources for discussion. In addition to documents such as the Interim Report for the project, and two proposals that had been drafted in connection with the Europe Code Week and the 'Widening Participation' initiative at Warwick, these included papers given just prior to C3 at the PPIG 2017 and ALT-C 2017 conferences in the UK, a paper to be given at <i>TAG in October and workshops for the Interactive Collaborative Learning 2017 and TAG conferences. A keynote paper for Edurobotics 2017 in November 2016 was also planned.

Making construals and coding

Given the prominence of 'coding' in current debates about computing and education, it was natural to revisit the relationship between making construals and coding. Hamish mentioned social science perspectives on computing, and referred to Peter Wright and John McCarthy's "Technology as Experience". For non-experts, it may be enough "to know what's possible ... what coding involves ... and not 'I can do it'". Markku expressed the need for something broader than coding. Peter remarked that programming cannot be subsumed into a broader non-technical activity 'some coding is always needed ... the threshold between coding and not coding is still there'. He also argued that where the skills behind coding are concerned: 'some people can't do things -- there are people who can't go through this'. Ilkka mentioned the role that unplugged activities can have in teaching basic principles, as when giving instructions to a blindfolded child navigating throgh the gym, and referred to the distinction between instructing a robot to move a step at a time and programming it to move: in step-by-step instruction "you don't need to think what will be the consequence". Peter referred to the essential characterisation of programming as algorithm + data, and commented that "you don't get this [the concept of data] from unplugged activities". To refer to data requires abstraction and variables, and this is a big step. Hamish likened this to "having difficulty with representation vs having difficulty with number". Peter referred to the difficulty, having observed a process in concrete terms, of then knowing 'how to write it down'. Conceptual difficulties included: when something repeatedly loops, recognising that something is the same, and of the limited semantics of the '=' sign. Ilkka mentioned the potential relevance of this discussion to the EU TACCLE 3 project [Estonia, Germany, Belgium, UK, Spain)].

The open online course (OOC)

Hamish reflected on the challenge of developing the open online course (OOC), especially where "we might have no idea who the audience might be", though this is not an unfamiliar problem for MOOCs (cf. http://www.research.ed.ac.uk/portal/en/publications/designing-for-the-unknown-learner(2dbf6edf-f078-4cd5-90c1-775933bdb28c).html and http://www.tandfonline.com/doi/pdf/10.1080/13562517.2015.1101680). The question of "What is the 'Hello World' of making construals?" remained for him potentially an important unanswered one. We could make a range of resources that can be tailored to many learners, for different purposes. An example might be aimed at primary school teachers, and a course based on the solar system. You might start with the [MCE] system and get the learner to explore the environment. Alternatively, you might present the system for the learner to deconstruct, as an exercise in comprehension. The aspiration is "for the learner with a blank sheet to do it themselves ... ." Where to put such a course? might use Moodle to colonise an institution (cf Russell Boyatt's proposal re Warwick), hoping to stimulate "open interactive activity". Could use a MOOC platform. Sustainability would be an important concern: would need to separate a developing course from the project deliverable (a copy archived on 31st August 2017). Functionality that would be needed from Moodle might include feedback questions and quizes. In addition, it would be important to consider the context for the OOC: whether you are freely giving access to other people, or facilitating in absence of other users. For instance, where anyone can download MOODLE and make an independent version the course is then 'out of your control'. One possible model is a "walled garden" supported by twitter communication or a slack group: this involves monitoring / moderating interaction among peers. Promoting interaction of this open nature between learners from many communities is relevant to particular subject matter: for instance, we might envisage a course on the Antikythera mechanis ranging from the mechanics perspective to astronomy, history, politics etc. In the school context, might need a single user copy of MOODLE (cf. a school may use Edmodo and provide the environment and links to support resources within it) .The open informal model of learner interaction is what is best-matched to the current form of the MCE. How far can we go towards a "tighter community as the next step"? For this we need more than help pages, must have a term, a course and a support community following a trajectory -- not a dialogic model for the OOC. Another possibility might be to follow a text (such as 'Children and Number: Difficulties in Learning Mathematics' by Martin Hughes)

The Making/Giving Change digression ... The 'Making Change' construal was discussed at this point (cf. the Skype discussion with Jane Waite at C6). This leads to a 'program' that incorporates a model of how the learner is thinking: translating human insight into an automatable procedure. Peter remarked that the main purpose of a PL isn't to translate human actions into a program but to translate knowledge from the person into maths/ In this sense programming is associated with modelling in an educational context (cf. David Riley) -- we "must instantiate the model in the computer: and this demands a specification". The construal of Giving Change "is much more difficult than modelling of change": so far from being simple, it is "demanding and more complicated".

Back to the OOC ... OOC curriculum would require a progression in content. This might begin with simple examples to motivate observables, dependency and agency, move on to definitions. (There are many different possibilities here - could have different urls for different level of difficulty. For instance: do we use functions or 'with'? might we able to avoid more complex functions, as in Peter's solar system model without sin() and cos()? If we used Antony Harfield's solar syste workshop as an introductory resource, we could implement proceduraul activity in many different ways (for instance, for managing time, use a clock mechanism, explicitly write "t++", use 'when' or use 'proc').

The MCE

The session ended with an extended review and discussion of the progresssion of models for managing scripts that had been introduced in the course of the proect when developing the MCE. There were broadly three generations of models to compare and contrast, as represented by construit.c1, the scifest variant (the 'classic' variant), and the redically revised construit.c6 variant. The way in which these handled different categories of definition in a script (such as extant, interpreted/recorded, as sequenced chronologically etc) were presented diagrammatically. Peter observed that in some respects the first model was the best!

The TPM M3 and practical in parallel sessions

Following the review, the management committee members participated in M3 (see the minutes in note form here), whilst others worked on resources. One remarkable product of the TPM session was the Factor Snake construal (originally proposed by Charles Crook) redeveloped in the MCE by Elizabeth Hudnott to demonstrate the process of incremental construction and the potential for adaptation. Meanwhile Nick Pope was working on an experimental mobile version of the MCE interface which made use of a more concise simplified syntax for with's and exploited SVG for visualisation. Some elements of this design were incorporated into later variants of the MCE.

Seminar at C3 Wednesday 14th September 2016 afternoon

The seminar was an 'edTech and Interactive Technologies' research seminar under the auspices of the IMPDET doctoral educational technology programme at UEF. It was introduced by Jaarko Suhonen: the timetable can be found at: https://wiki.uef.fi/pages/viewpage.action?pageId=43649511.

The first talk, given by Ilkka Jormanainen introduced the CONSTRUIT! project, and related it to the TACCLE3 Erasmus+ projects under the banner of "KA2 cooperation for innovation and the exchange of good practices". After framing the concept of CONSTRUIT! as "enabling educators and learners" to collaborate in creating live interactive resources ("construals") and describing the suite of activities (from building, to understanding a ready-made model and modifying it, and to using a model), he introduced the MCE as one environment that unified such activities and was intended to promote "Making Construals Everywhere". Meurig Beynon gave a short informal presentation to illustrate making construals in the MCE at this point. This was based on the workshop session at ALT-C just prior to the C3 meeting, and incorporated a demonstration of the solar system construal and the construal of the recession of the planets as revealed in the earth-centric view. The seminar session was being relayed online and Chronis Kynigos tweeted his appreciation from Athens.

The second talk, given by Tapani Toivonen, outlined his research on applying the "Empirical Modelling (EM)" principles behind making construals to unsupervised m/c learning associated with the INternet of Things (IoT). The aspiration in this work was to build on prototype applications devised by Ilkka Jormanainen in his PhD that used the EDEN interpreter to analyse data interactively in Observable-Dependency-Agency terms. The basis for Tapani's innovation was a "whole new approach to clustering (in field of educational technology)" in which the challenges to be met included large volumes of data, preparation of data sets and inconsistent representations. These principles were also to be applied in a Horizon 2020 eCraft2Learn project that had recently been awarded to UEF.

The third talk, give by Solomon Sunday Oyelere from Nigeria, a PhD student at UEF, took the form of an evaluation report on a mobile learning supported classroom. This exploited small learning apps on a mobile phone which promoted social networking and blended learning.

Final sessions: Thursday 25th September morning

The penultimate session of C3 was devoted to two themes relating to making construals.

The first was a presentation from Markku Tukiainen devoted to his PhD work on spreadsheet programming (see ftp://cs.uef.fi/pub/Dissertations/tukiainen.pdf). This presentation underlined the important distincion between the use of dependencies in the spreadsheet grid, where a structure has to be imposed on the observables associated with the cells, and the definitive scripts recorded in the MCE, which can be freely organised. Following principles of programming based on Elliot Soloway's 'goals and plans' it is possible to organise computations in a spreadsheet in such a way that their structure reflects goal-oriented constructs. The tool devised for this purpose is the 'structural spreadsheet experimental tool' BASSET. Some thought-provoking ideas that may be relevant to making construals stem from this work. Practices followed in building scripts from scratch resonate with the claim that 'spreadsheets can be 'too easy to use' in the sense that 'reference in the grid can be too random and disorganised, even chaotic'. Creating mature construals that are relatively modest in size compared with the 'spreadsheets with about 20K cells' analysed in Markku's thesis points to the need for some methods of organising definitions to aid comprehension. On this account, some ways to structure are necessary, but it also appears that this should be fluid and pragmatically determined. The fact that spreadsheet users can't adapt their old manual ways to use BASSET effectively is consistent with empirical evidence from working with construals which suggests that imposing fixed structures on definitive scripts in the MCE can be obstructive. A more fundamental connection between making construals and Soloway's principles of programming is suggested by the title of his paper "LEARNING TO PROGRAM = LEARNING TO CONSTRUCT MECHANISMS AND EXPLANATIONS" (see http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.11.6583&rep=rep1&type=pdf), as a construal is associated with both 'mechanisms' and 'explanations'.

The second brief presentation, given by Ilkka Jormanainen, discussed possible connections between CONSTRUIT! and the TACCLE3 agenda. Partners within this consortium, which embraces Tallinn / Pontydysgu / Karlsruhe (Kit) / Estonia (Hitsa) / Salamanca / Antwerp (Belgian shcools) / Aalto (Mario Virnes), UEF, has generated numerous lesson plans which might be of interest. The materials being generated are mainly static resources, such as text and pictures, and there might be an effective way to develop more interesting 'unplugged' and interactive resources through making construals.

he final session of C3 considered some practical issues concerning the content of the construit.org website.

Ilkka Jormanainen and Rene Alimisi reviewed a worked example on the site which unfortunately, but unsurprisingly, was now related to an obsolete variant of JS-Eden. We considered the form that example construals on the website might take: which to choose and what worksheets or other documentation to attach to them. Some prototyping of a possible illustrative example was explored collaboratively. This exercise highlighted problems already encountered in devising and presenting introductory resources for the MCE: - the challenge of the proposed 'walkthrough' construals was revisited (these related to the problems of animating interaction on screens with different resolutions) - the nature of introductory vides and presentations was discussed: we looked at a video and JSPE presentation developed by Jonny Foss for the Warwick Postgraduate Award in Technology Enhanced Learning (PGA TEL) course. Jonny joined us in this discussion via Skype. Another topical isue for UEF was what examples of making construals might be suitable for a Forum such as SciKids -- possible candidates included games such as XXXI and card games such as 'diminishing whist' and 'patience'].

Reflection on "What is distinctive about construals?" caused some controversy. Nick Pope challenged "false claims" about the ease of making construals rather than 'writing programs'. Teachers with trad programing experience would disagree with claims that changing parameters such as the number of moles and the nature of the images etc in Whack-a-Mole were particularly easy -- they can after all typically be done trivially by modifying program code, whatever the programming language, and can be done in a very similar live editing style in any contemporary development environments. Likewise the appeal to flexibility can be challenged: many constructions are quite easy to program from scratch -- why bother to make or revise an existing construal to achieve a specific goal? To make the case for making construals as an 'easier' way of programming requires better examples that are not as simple as the solar system construal and other toy educational examples. These challenges underline the need to take care in framing the case for making construals, and putting the emphasis on qualities that have more to do with semantics, pedagogy and epistemology than simply developing programs to meet a functional specification.