Intelligent Physical Blocks for Introducing

Computer Programming in Developing

Countries

Andrew Cyrus Smith

African Advanced Institute for Information & Communications Technology, Meyring Naude
Road, Brummeria, Pretoria, South Africa.
Tel: +27 12 8414626, Fax: + 27 12 8414720, Email: [email protected]
Abstract: This paper reports on the evaluation of a novel affordable system that
incorporates intelligent physical blocks to introduce illiterate children in developing
countries to the logical thinking process required in computer programming. Both
the usability and educational aspects are reported on. We provide a brief overview on
previous work in this field. Results obtained from field studies are given. We
conclude with recommendations for improvements and further research.
Keywords: GameBlocks, illiterate, programming, physical blocks, education,
usability, developing countries, pre-school.

1. Introduction
Perhaps the greatest stumbling block to entry into the domain of Information and
Communication Technologies (ICTs) among young people in developing countries is the
lack of informal exposure to personal computers. It can be argued that formal exposure to
ICT is not ideal in all current education systems. Some current education systems are biased
in favour of good/photographic memory abilities, allowing a candidate to pass a subject
without a true understanding of the contents. A better system could be the informal
exposure to the learning subject, outside of the classroom, making it part of the childs
recreation activities instead of a formal learning experience. It has previously been reported
that children learn better when they are exposed to fun activities, rather than as a formal
lesson.
Three-dimensional vision is learned [1, p51], enabling us to perceive it through twodimensional media such as written text. Hannaford calls this visual literacy. People who
have not yet mastered it are unable to interpret a drawing of a mountain as such. They only
notice the colours and lines on a piece of canvas, but not the texture or perspective.
Learning is done best when multiple sensors are involved. Learning through reading twodimensional text or listening to a teacher limits the learning experience. In contrast,
learning through experience combines senses, emotions, and movements that involve the
learner fully [1, p5]. Touch, specifically, is a major contributor to fully understanding in
vision. [1, p50] Our system exploits this knowledge about learning by providing a tangible
interface which controls another tangible object. This helps make the learning concrete.
We support the thinking of others [2, p3][3, p364] who question whether perhaps the
way ICT is taught should be adapted to the local context.
The system reported on here is aimed at introducing young children to programming
where it is more important to gain experience with simple programmes, rather than more
complex ones [4, p20]. Its aim is to introduce young children to the basic logical thinking
required for writing computer programmes, but without the requirement of first learning to

read and write, operate a computer, using a computer keyboard, typing, saving the code,
compiling it, downloading, and then executing the instructions. The system aims to bypass
all these literary requirements, making a direct link from logical thought to physical
execution. This is done using simple electronic circuitry, removing the requirement for a
traditional and relatively expensive, fragile computer in the form we have come to know as
the Personal Computer. This makes it particularly suitable for developing countries due to
its affordability.
Our approach is supported by Papert [5, p23] where he states It is not necessary to
work with computers in order to acquire good strategies for learning.
It can also be argued that the system supports kinaesthetic thinkers [4, p17] better
than traditional keyboard-based programming in exercising the logical thinking process.
We admit that there are currently limitations to the system. For instance there is no
simple way to restore a previously-coded programme, or implementing complex coding
structures. But after all, the aim of this system is to introduce young children to the creative
and logical process of simple programming.

2. Prior Work
Other systems using physical objects have previously been developed
[3][4][6][7][8][9][10][11], either using electronic components inside the objects
themselves, or using optical cameras to read identifying information off the objects. Our
system offers the distinct advantage of reducing the cost of the blocks significantly in that
only magnets are used inside the blocks to represent their functionality, albeit with reduced
functionality. We believe that this approach is more appropriate for developing countries
because of the resultant significant monetary savings.

3. Objectives
We tested the GameBlocks [14][15] systems usability and educational value when used to
teach illiterate children the principles of basic programming.
The usability outcomes evaluated, based on ISO 9241-11 [12], are firstly the
effectiveness (degree of task completion), secondly the efficiency (how much effort it takes
to complete a task), and thirdly the satisfaction (positive attitudes toward the product and
freedom from discomfort) in which the user can use the system to achieve a pre-determined
goal.
The educational outcomes that were evaluated are the ability of the testees to firstly
compile a sequence of instructions in the correct order from those available, secondly the
correct translation by the testees of the written instructions into the symbols on the cubes,
thirdly the testees placing the cubes onto the trays in the correct sequence, and lastly to
understand how the instructions represented by the cubes are transferred to the toy robot
[13].

4. Technology Description
GameBlocks is a system consisting of acrylic blocks (Figure 1), trays (Figure 2), and
controlling electronics that enables an illiterate child to compile simple physical sequences
to control a toy robot (Figure 3). Each 3-dimensional block is colour-coded and has a
symbol on its top surface. Together, the colour and the symbol both represent the function
the toy robot would execute. The user chooses, from the available blocks, the sequence of
movements the toy robot has to execute, and places them in that same sequence on the trays
(Figures 4,5). As the electronic circuit interrogates each block sequentially, an appropriate
command is sent to the toy robot for interpretation and immediate execution (Figure 6).

Figure 1. Acrylic block.

Figure 2. Tray.

Figure 3. Humanoid toy robot.

Figure 4 (left) and Figure 5 (right). Children placing the blocks on to the trays.

Figure 6. (top)Physical sequencing (left to right) example of 7 blocks as viewed from above. (bottom) The
resultant movement over time (left=start, right=end) of the toy robot as viewed from above.

5. Methodology
Multiple workshops were held at two science events, with groups of approximately 20
children participating in each workshop. Some of the workshops were formally evaluated.
During the formal evaluation in the order of 30 children participated. During the informal
evaluation sessions approximately 50 children were observed.
5.1 Formal evaluation
A company external to our institute was contracted to provide an independent report on the
system and the childrens interaction with it. The workshop was run in a venue normally
used for dancing rehearsals, with a wooden floor and the walls and ceiling painted black.
One wall was completely covered by a mirror. Supporting project team members were
present during the tests, positioned behind the testees and having a clear view of the
activities. Verbal comments given by the supporting team members after each experiment
served to inform this paper and the resultant formal report [13]. Although written
instructions were used in the tests, we believe that the outcomes would be no different if
only symbols had been used instead.

5.2 Preparation
Prior to each evaluation, four robot movements were written on strips of white paper
approximately 30x15cm in size (Figure 7), one movement per strip. The wording on the
strips were: Forward, Backwards, Left, and Right.

Figure 7. Pseudo- code strips.

Figure 8. Symbols used for evaluation.

These represented forward, backward, left, and right toy robot body motions. Although six
toy robot movements are supported by the system, only these four were made available for
the evaluation. Two acrylic blocks and two paper strips for each movement instruction were
used in the test, allowing for a total of eight possible instructions.
Also on separate paper strips, approximately 15cmx15cm in size, were drawn happy and
unhappy faces (Figure 8). These would later be used to denote the success or failure of a
step in the test. Prior to the testing these paper strips were randomly stuck to the nearby
wall using sticky putty (Figure 7). The sticky putty allowed for easy repositioning of the
paper strips later.
The trays were placed side-by-side in a singular row, and the blocks interpreted
sequentially from left to right as faced by the testee (Figure 6). Once the testee had
completed placing the blocks onto the trays, a verbal indication was given to the facilitator.
Interpretation by the electronic circuitry commenced when the facilitator switched the
system on. Alternatively, the system makes provision for immediate execution when a
block is placed onto a tray, but this mode of operation was not evaluated.
Both still- and video-cameras were utilised during the sessions to aid in later analysis of
the tests. The two camera operators would roam the test area so as to capture images from
the best vantage point.

Figure 9. A tester takes notes (to the right at a table) while the facilitator (to the left) assists atesteer (centre)
in deciding where to place a block onto the trays with other seated testees waiting their turn (foreground).

On the floor, in front of where the testees were seated in a semi-circle facing the
instructor, were the trays, blocks and controlling electronics (Figure 9). The trays were
connected to the controlling electronics with wires. The blocks were randomly positioned
together on one side of the set-up, with no particular meaning in their configuration.
However, they were all orientated so that the symbols on the top of the blocks were facing
upwards. Some blocks were stacked on top of others.
5.3 Reception and privacy
There was no pre-selection possible during these events. Children arrived by bus from
neighbouring areas and attended the workshops without having had any a-priori contact
with the researchers. In each instance the group would comprise children from a single
school, accompanied by an adult who took responsibility for them.
Upon arrival at the workshop entrance, the following formalities were taken care of.
Firstly a consent form was required, signed by the adult accompanying the children. This
form included information about the purpose of the experiment, and the way in which the
childs rights would be protected. For example, the childs identity would not be made
known in subsequent reports. Identity includes the childs name and face. It was also stated
that any published results would not be connected to the individual and the individual
would not be identifiable by the results. This consent was required for the group and not per
individual in the group. Secondly each child had a large (approximately 10cmx10cm) white
paper label with the childs age pinned to the chest region (Figure 10). This visual
information could later be used in analysing the video footage.

Figure 10. A tag with the users age would later aid in the analysis of the video material.

5.4 Measurement
The testee first identified the instruction (pseudo-code) to be executed. These were then
placed on the wall in the desired sequence, visible to all testees (Figure 11). When satisfied
with the sequencing, the testee placed a corresponding block on a tray (Figures 4,5,9).
When satisfied that all the blocks have been placed in the required sequence as indicated by
the pseudo-code, the testee signalled the facilitator. The facilitator then activated the
electronic circuitry and the commands were sent sequentially to the toy robot for execution.
As each instruction was executed, the help of all testees was elicited to verify that the
movement corresponded to the pseudo-code. If it did, a paper strip with a smiling face was

positioned below the relevant instruction on the wall. If it did not, an unhappy face was
used instead. This was later used to determine the effectiveness of the system.

Figure 11. A user(right) places pseudo code in the desired sequence on the wall.

5.5 Instructions
English was used as the instruction medium. Although some testees had another mother
tongue, they could all converse in English to varying degrees. Having taken their positions
in no particular order around the already set up equipment, they were welcomed by the test
supervisor team member. At this point they were briefly introduced to the project and the
purpose of the test. It was explained that they were participating in a test to measure the
usability- and educational-value of the GameBlocks, that there was no right or wrong way
to participate, that their inputs would be considered to improve on the current design, and
that their identity would be protected. They were informed that a video recording as well as
still photo shots would be taken during the session.
The tester then introduced the technical expert who managed the rest of the experiment.
The tester, camera operators, and rest of the team members observed the proceedings that
followed in silence from a distance.
The approach taken by the technical expert was very different to that of the tester. The
expert provided another perspective on the project to the children, explaining that the aim
was to develop an interest in science and technology amongst young school-going children.
At this stage the various components of the system were introduced.
In both cases the test supervisor and technical expert kept the introduction to the system
brief, pointing out the system components to the testees and giving a short verbal
explanation on how to use it. This was done in order to evaluate the intuitiveness of the
system.

6. Results
6.1 Usability results
Effectiveness The brief introduction about the system proved to be insufficient for the
children to grasp its purpose and usage. Multiple demonstrations were required before they
were successful at using the system. This could be ascribed to a number of reasons. Firstly,
the symbols on top of the blocks were not intuitive; the testees did not understand them
without assistance from the instructor. Secondly, the required orientation in which the
blocks had to be placed onto the trays was not well understood. Thirdly, although the
mechanical design of the system ensures that the block will fit squarely onto the tray, it was
not clear to the testees which side should face up. Fourthly, there was some confusion

regarding the orientation of the symbol and the resultant movement of the toy robot. An
explanation could be that the symbol represents the movement relative to the toy robots
own co-ordinates, but the testees expected the movement to be relative to the rooms coordinates. After the first testee had finished the test, the subsequent testees who had
observed the first were more successful in completing their tasks [13].
Efficiency The first testee spent close to 15 minutes mastering the use of the system to a
sufficient degree to complete the set task. The second and third testees, who had observed
the first, showed an average improvement of 60% in completing the task [13].
Satisfaction Feedback received from the testees indicates that they had enjoyed the
challenge and the ability to control a toy robot. Some changes were recommended: the
movements made by the robot were not visible enough and could be accentuated; colourcoding of the symbols would be useful (the cubes were already colour-coded); the
movement directions represented by the symbols should be relative to the room and not the
dynamic computational style as described in [5, p55] (for instance, if the symbol indicates a
movement in the direction of a window in the room; the toy robot should move in that
direction); smaller blocks would be preferred; the mechanical reliability of the concept
demonstrator was lacking, resulting in unreliable operation; the speech produced by the toy
robot was inaudible [13].
6.2 Educational results
In all tests the testees were eventually able to correctly sequence the blocks onto the trays
according to the pseudo-code. In one test the first testee placed the blocks onto the trays in a
random fashion, seemingly unaware that it had to be done according to the pseudo-code.
This testee managed to correct the sequence after intervention by the instructor.
All testees were able to volunteer plausible mechanisms regarding the communication
method employed between the system and the toy robot. Through conducting interactive
group discussions with the testees, we elicited from them the method the system
communicated with the toy robot, eventually correctly concluding that the actual method
was infra-red emission [13].

7. Conclusions
Our goal was to design a Physical User Interface (PUI) to explore alternative ways to
introduce young illiterate children to the art and science of computer programming in
developing countries where informal exposure to computer programming is not readily
available. It is hoped that this interface would also address the declining physical activity of
those children who have unchecked access to modern recreational computer-based games.
By requiring physical movement of the whole body when using the PUI, and not just the
fingers, GameBlocks has the potential to also appeal to illiterate pre-school children whose
fine motor skills are still developing..
We reported on the usability and educational value of a novel affordable physical
programming system. This was achieved without the use of an expensive computer. Tests
indicate that significant learning is required for a novice user of the current GameBlocks
implementation. However, all testees were able to order the instructions sequentially and
interpret the symbols on the blocks to match the instructions and then place the blocks in
the correct sequence onto the trays. We observed that learning did not take place primarily
as a result of self-confident exploration or the intuitive-ness of the design, but rather mostly
because of the interventions made by the facilitator. This new type of game was successful
in creating great excitement amongst the testees, but the disadvantage was the uncertainty
exhibited by them and the subsequent prolonged learning timeline.

We believe that our system also addresses the challenge of introducing a novice to the
lower-level of abstraction [16, p70] as it relates to computer programming because there is
a one-to-one correlation between the instructions and the resulting movement of the toy
robot.
Similar to the aims of others when setting up a learning environment for using
computers [17, p 43], our system-design facilitates movement and action around the blocks,
reinforcing communication and information sharing amongst the children using it.
Recommendations
Firstly, the co-ordinates used by the toy robot could be made more explicit by improving
the symbology used on the blocks (Figure 12). This would potentially resolve the confusion
experienced by the testees who assumed room co-ordinates in their thinking processes when
in fact the toy robot used its own co-ordinates when executing instructions [13]. Secondly, a
mechanical linking mechanism between the blocks could be introduced to make sequencing
more obvious to the first-time user (Figure 13). Thirdly, adding distinct markings on the
front and bottom of the cubes could help overcome the problem with orientation which
some testees experienced.

(a)
(b)
(c)
(d)
Figure 12. Proposed improved block symbology, based on [13]. (a)Move to the left. (b)Move to the right.
(c)Move forwards. (d)Move backwards.

Figure 13. Proposed improved mechanical orientation mechanism, based on [13].

Future research
Further research is required for the development of an appropriate physical and visual
syntax to represent more complex coding structures. Extended formal tests on pre-selected
participants are needed to determine at which developmental stage the child will benefit the
most from a system of this kind.
Tests with cubes of various sizes should be conducted to confirm the most ageappropriate block configuration.
Future research will explore the use of physical artefacts that are well-known to the
target group, that is, children in developing countries. As an example, instead of using

synthetic materials such as acrylic in the current system, wooden blocks carved by local
craftspeople could be utilised. This idea is supported by [2, p3] where the author states that
A particular educational principle can well be transferred to another setting but it should
be modified to accommodate the new cultural setting. (emphasis added).
Commercialisation
A number of issues have to be resolved before the GameBlocks system can be
commercialised. These include the optimisation of the electronic circuit design, determining
the optimal target market, developing clear symbology for the blocks, expanding the current
instruction set, and using a child-appropriate material for the blocks.

Acknowledgements
We wish to thank Mpho Nkuna for implementing the design and Corne Kruger for
designing and executing the usability tests. The FabLab at the Pretoria Innovation Hub
manufactured acrylic trays and blocks. Reinett Mafentile assisted with assembly. Feedback
received from the conference reviewers is greatly appreciated. This research was funded by
the South African Department of Science and Technology.

References
[1] C. Hannaford, Smart Moves: Why learning is not all in your head, 2005.
[2] M.C. Grove, H.M.A.M. Hauptfleisch, Stepping stones: A school readiness programme, Part 1 Teachers
Manual. De Jager-Haum, ISBN 0-7986-1488-9, 1985.
[3] M.O. Vesisenaho, H.H. Lund, (2004) I-BLOCKS for ICT Education Development, Case Iringa, Tanzania.
In Proceedings of the 33. International Symposium IGIP / IEEE / ASEE, September 27-30, 2004, Fribourg,
Switzerland, University of Applied Sciences of Western Switzerland, pp. 364 - 371.
[4] T.S. McNerney, Tangible programming bricks: an approach to making programming accessible to
everyone. Program in Media Arts and Sciences, MIT, February 2000.
[5] S. Papert, Mindstorms, Children, computers and powerful ideas, Basic Books, Inc. 1980.
[6] A.F. Blackwell, R. Hague, Autohan: An architecture for programming in the home. In Proc. IEEE
Symposia on Human-Centric Computing Languages and Environments 2001, pp. 150-157.
[7] T.S. McNerney, From turtles to Tangible Programming Bricks: explorations in physical language design.
Personal Ubiquitous Computing, 8(5), Springer-Verlag (2004), pp. 326337.
[8] H. Suzuki, H. Kato, Interaction-level support for collaborative learning: Algoblockan open programming
language. In Proc. CSCL 1995, Lawrence Erlbaum (1995).
[9] A. Valente, Explorations in theoretical computer science for kids (using paper toys), In Proc. IDC 2004,
ACM Press.
[10] P. Wyeth, H.C. Purchase, Tangible programming elements for young children. Ext. Abstracts CHI 2002,
ACM Press (2002), pp. 774775.
[11] J. Nielsen, H.H. Lund, Contextualised Design of African I-BLOCKS, Fourth IEEE International
Workshop on Technology for Education in Developing Countries, pp. 14-18, 2006.
[12] International Standards Organisation. 1998a. ISO 9241-11: Ergonomic requirements for office work with
visual display terminals (VDTs) Guidance on usability. ISO: Geneva, Switzerland.
[13] A. Bekker, C. Kruger, Usability and educational review on GameBlocks and Body Ping Pong. Test
report, Test and Data Services, Available online
http://playpen.icomtek.csir.co.za/~acdc/education/ScienceUnlimited2006/reports/ICT in Education Report
1.0.doc , 26 April 2006. Accessed 5 January 2007.
[14] A.C. Smith, Tangible cubes as programming objects. In Proc. ICAT 2006, ISBN-13: 978-0-7695-2754-3,
IEEE Computer Society (2006) Order Number P2754, pp. 157-161.
[15] A.C. Smith, Using Magnets in Physical Blocks That Behave As Programming Objects. In Proc. First
International Conference on Tangible and Embedded Interaction 2007, ACM Library (2007). To be published.
[16] I. Jormanainen, O. Kannusmaki, E. Sutinen, (2002). IPPE - How to Visualize Programming with Robots.
In M. Ben-Ari (Ed.), Second Program Visualization Workshop (pp. 70-74). University of Aarhus, Department
of Computer Science.
[17] I. Harel, S. Papert, (Eds.), Constructionism. Ablex publishing corporation, 1993.