Five Teams, One Bridge, and the Joint That Has to Close
In Singapore the large collaborative construction sits in an odd corner of the corporate catalogue: roughly three operators publish it as a named product, it is absent from almost every general activity round-up that HR teams actually search, and what is published about it is thinner than for any other format at a comparable price. The listings say a group of somewhere between eight and five hundred people will spend one to three hours building something together, indoors or outdoors, and that the pieces will combine into a city, a kingdom, a landmark or a bridge. What none of them tells you is the single detail on which the whole session turns, namely how the pieces are meant to meet.
That is the whole activity. Each team builds one component in isolation, and at the end the components are brought together and have to physically fit. Not sit next to each other. Fit. A deck has to align with the next deck, a road has to arrive at the district boundary at the same width it left the last one, a wall has to meet a wall. This page is written so that an organiser can specify that fit properly, choose a material that makes it possible, decide how much the teams may talk to each other, schedule an assembly instead of hoping for one, and score the result in a way that stops the most impressive component from winning.
A one line answer on whether PLAYON supplies this
PLAYON does not run collaborative construction sessions: there is no build space, no materials store and no facilitated construction product on the floor, so if this is the format you want, brief a corporate build supplier or run it yourself. Everything below is about running it well.
How this format is packaged and priced in Singapore
A collaborative construction is a build challenge in which a single large structure is divided into components, each component is assigned to one team, the teams build independently, and the components are then assembled into one object that must hold together and usually must pass some kind of test. The common briefs are a model city divided into districts or zones, a fantasy kingdom divided into quarters, a landmark divided into storeys, and a bridge divided into spans.
Three things are worth extracting from what local and regional operators actually publish.
The published group bands are enormous and almost useless on their own. One local page publishes 24 participants to unlimited. Another publishes 8 to 500 across an entire shelf of build products. A third publishes tiered bands that step from 26 up through 50, 150, 250 and 500 participants, with an open band above that. Nobody publishes a component count against a headcount, which is the number you actually need, and it is worked out further down this page.
Published durations cluster at one to three hours. One local listing publishes 1 to 3 hours, another publishes fixed options of 2, 2.5 and 3 hours. Read those carefully, because a two hour figure that includes a brief, a build, an assembly and a debrief is a very different product from a two hour build window with everything else on top. When you compare two proposals, ask for the assembly minutes separately. That single question sorts suppliers faster than anything else.
Nobody publishes a per person rate for it. Not one of the pages read for this article carried a price. So the only honest guide is the general market. In your own office with no catering, a single facilitated activity is published at $25 to $40 a head. A facilitated half day in a hired venue with light refreshments is published at $45 to $90. At the premium full day tier, food and a managed room included, the published figure is $100 to $180. Venue hire on its own is published separately at $10 to $40 a head. Surcharges published across the market are 15 to 25 per cent on a weekend and about 30 per cent on a public holiday, and food bundled into an all-in figure carries a published uplift of 30 to 40 per cent. A facilitated build with supplied materials, a judging panel and an assembly crew belongs in the second of those tiers for most groups.
One more observation that matters more than any of the numbers. Of the local round-up articles that list twenty or more indoor corporate activities, most carry no build format at all. The family is genuinely uncommon here, and the uncommonness is not a verdict on the format. It is a verdict on how hard it is to deliver: it needs floor area, a materials budget, an assembly plan and a facilitator willing to let a room fail in public. It is easier to sell an escape room.
Sequential failure, invisible seams, and the one that has to physically fit
Three formats in this corner of the catalogue look similar on a proposal and behave completely differently in a room. Getting this distinction right before you brief anybody is the most valuable ten minutes you will spend, because two of the three are cheaper and easier and neither of them does what this one does.
The chain reaction build passes energy along a line. Each team builds a segment, the segments are lined up, and one trigger sets the whole thing off. The dependency is sequential and the failure is total: if the third join does not transmit, nothing downstream of it ever moves, and five teams watch their working segments sit there unused. What crosses the boundary is motion, not matter. Two segments never have to be the same height, only close enough for a marble to make the trip.
The collaborative canvas splits an image. Each team paints one panel of a picture nobody has seen whole, and the panels are laid out together at the end. The split is visual and there is no physical join at all: the panels touch but nothing passes between them and nothing bears on anything. Crucially, a mismatch is soft. If the ochre band arrives two centimetres too low, the assembled piece reads as slightly odd, most of the room does not notice, and a generous eye calls it handmade. The seam can be absorbed.
This format has to physically fit, and that changes everything. The components are matter. They occupy space, they have dimensions, they bear load, and at the end of the session two of them are pushed together in front of everyone. There is no version of a mismatch that reads as texture. A deck built at 22 centimetres and a deck built at 15 centimetres do not meet, and no amount of goodwill in the room makes them meet. Someone has to cut something, or the structure does not exist.
Two consequences fall out of that, and they are the reason to choose this format over its two neighbours.
The specification of the interface becomes the entire exercise. In a chain reaction the interface is a question of energy transfer and a generous receiving tolerance solves most of it. In a canvas the interface is a line on a printed edge and a five minute conversation solves most of it. Here the interface is a set of physical dimensions that have to be agreed, written down, measured and held by five independent teams working out of sight of each other for ninety minutes. That is a specification problem, and it is the same specification problem that every organisation with two teams building against a shared boundary already has.
The mismatch cannot be hidden, and that is the point. The canvas forgives. The bridge does not. When the fifth span is carried across the room and set down against the fourth, the room can see instantly whether the two teams were building the same thing. There is no interpretation, no debate and no facilitator judgement. It either meets or it does not, and everybody watching knows which. Very few corporate exercises produce a verdict that honest, and the honesty is the whole value.
The bridge brief is the sharpest version of the family for exactly this reason, and it is the one to run if the group has a real handover problem. A city brief is more forgiving, because a district that is three centimetres out can be nudged, and a kingdom brief is the most forgiving of all, because fantasy architecture absorbs almost any discrepancy. If you want the lesson to land, take the forgiveness away.
Step by step, from arrival to the last connection
What follows is the order of play for a session you run yourself. It assumes you have chosen a brief, a material and a component count, all of which are covered below.
Mark the assembly footprint on the floor before anyone arrives. Tape the outline of the finished structure, with the component boundaries marked and numbered. Teams should be able to walk over and look at the space their component has to occupy on the day, because a footprint on the floor is a specification everybody can see.
Brief the whole room together for fifteen minutes. Cover the finished object, the fact that everything will be assembled and tested at the end, the scoring weights, the safety rules for blades and glue, and the central specification. Say plainly which parts of the specification are fixed and which parts the teams have to settle between themselves.
Issue the written specification on paper, one copy per team, and read it aloud. A specification that lives only inside a spoken brief will be remembered five different ways by five different teams.
Run the specification conference as a timed block. Ten to fifteen minutes, one delegate per team, standing at the taped footprint, settling everything the central brief deliberately left open. Nothing else in the day carries as much weight as these fifteen minutes. The method is set out below.
Write the agreed specification on a flipchart, in numbers, and photograph it. Every team gets the photograph. An agreement that is not written in millimetres is not an agreement.
Issue identical material kits and confirm component assignments. Identical kits are not a nicety. If one team has more board than another, the debrief becomes an argument about fairness instead of a conversation about interfaces.
Open the build window of sixty to ninety minutes. Teams design and build inside their own component. Call the remaining time at the halfway mark, at thirty minutes, at ten and at two.
Run the scheduled communication windows. These are announced blocks in which teams may contact their neighbours. How many windows, how long, and how tightly restricted is your largest single design decision, and it gets a section of its own further down.
Close the build and call the freeze. Nothing is cut, glued or added after this point. Tools down, and mean it, because a team still cutting during assembly is a team that will hand you a component with wet glue.
Move the components to the assembly footprint, in order, by a named crew. Not by everybody. Two or three people carry, the rest stand behind the line.
Assemble, on the clock, in a stated order. For a bridge, from both ends inward, so that the last joint to close is the middle one and the accumulated error lands where the room can see it. For a city, from a fixed central datum outward.
Run the test. A weighted object driven or dragged the length of the deck, a load placed on the roof, a ball rolled along a street. State the test in the opening brief so teams build for it.
Name where it failed, precisely and out loud. Inside a component, or at a joint, and which joint. Do not soften this and do not skip it. The location of the failure is the material the debrief runs on.
Open a repair window of ten to fifteen minutes, then test again. Almost every organiser should plan a second attempt.
Walk the structure for the component presentations. Sixty to ninety seconds per team at their own component.
Debrief within fifteen minutes, in the room, seated, before anyone leaves.
Writing the brief: what the centre fixes and what the teams must settle
Everything difficult about this format is contained in one question. How much do you specify centrally, and how much do you leave to the teams?
It is a genuine fork with a bad answer at each end, and organisers usually arrive at one of the two bad answers by accident rather than by choice.
Specify everything and you have run a kit assembly. Issue every dimension, every connection method, every height and every tolerance, and the components will fit. Of course they will. Five teams following five sets of complete instructions produce five parts that mate, and the assembly is a formality. What has been exercised is following instructions carefully, which is a real skill and is not what anybody booked a room for. The hardest problem in the session was solved by an authority figure before the teams met each other, and the debrief has no raw material at all, since the truthful reply to what did your team decide is that the paperwork decided it.
Specify nothing and you have run a demonstration of futility. Tell five teams to build a bridge span each and say nothing further, and you will get five spans at five heights, five widths and five connection philosophies, discovered at minute one hundred and ten with no time to repair anything. The lesson lands as arbitrary bad luck rather than as insight, and a room that feels tricked does not debrief well.
The resolution is to fix the envelope centrally and leave the interface to the teams. Fix the things that are not negotiable and that no team could reasonably discover, then deliberately withhold the things two neighbours have to agree.
What the centre should always fix:
The finished object and its overall dimensions. A bridge with a total span of six metres. A city occupying a three by four metre footprint. This is not negotiable because it is determined by your floor, not by anybody's preference.
The component count and each component's slot. Five spans of 1.2 metres, numbered one to five, in a stated left to right order. Teams need to know who their neighbours are from minute one.
The materials and the quantity. Identical kits, stated on paper.
The test. What the finished structure has to do. A one kilogram weight travelling the full length of the deck. A load supported at the centre for ten seconds. State it precisely, because a vague test is an argument waiting to happen.
The clock and the communication rules.
Safety limits. No standing on furniture, blade rules, glue rules, and a maximum height nobody may exceed.
What the centre should deliberately leave blank:
The mating dimensions. Deck height above the floor, deck width, road width, wall height, roof line. This is the blank that does the work.
The datum. Where zero is measured from.
The connection method. How two components are actually joined at the end.
The tolerance, and who absorbs it.
The internal design of each component.
That list of blanks is the exercise. The teams have to notice that they are blanks, work out which ones matter, agree numbers, write them down, and then hold to them for ninety minutes while nobody is checking.
Six things a fit specification has to name
Give the teams these six headings in the brief. Name them, do not fill them in. A specification that covers all six will hold. One that misses any of them will fail at exactly the missing clause, and the missing clause is almost always the first one.
The datum. Where zero is. This is the clause nobody thinks of and the one that ruins the most sessions. If team two measures deck height from their table top and team three measures from the floor, and the tables are 74 centimetres high, the two decks are three quarters of a metre apart and both teams built exactly to the agreed number. Fix the datum out loud, in the specification conference, before any dimension is discussed: all heights are measured from the finished floor, at the taped footprint, with a tape measure that stays with the specification keeper.
The mating dimensions. The actual numbers at the boundary. Deck height 150 millimetres. Deck width 200 millimetres. Kerb height 20 millimetres. Written in millimetres, not in words, and never as "about a hand's height", which is the phrase that produced the failure in the worked example further down.
The joining face and its orientation. Which face of your component is the one that meets the neighbour, whether it is vertical, and whether it is square to the line of the structure. A span that arrives at the right height but leaning six degrees off vertical will not sit against its neighbour.
The connection method. How the two components are physically fixed together at assembly. Butted and taped. A tab that slots into a matching slot. An overlap plate that spans both. A dowel through matching holes. Whichever it is, both sides have to build for it, and a team that builds a tab against a neighbour who built another tab has two tabs and no slot.
The tolerance, and who takes it up. How much slack is allowed at the boundary, and whose job it is to absorb the difference. The strongest single clause available to a group is a stated adaptor allowance: each component is built 10 millimetres short of its nominal length, and the gap at every joint is closed by a small connector piece built during assembly. That one clause converts a precision problem into a tolerance problem and saves more structures than anything else on this list.
The load at the joint. What has to travel across or bear on the boundary. A joint that only has to look continuous is a different object from a joint that has to carry a rolling weight, and teams that have not been told which one they are building will build the first and be tested on the second.
Deliberate under-specification, and the bridge as the sharpest case
There is a version of this session in which you leave the blanks not because you forgot but because the blanks are the product. It is worth naming as a deliberate design choice, because facilitators drift into it accidentally and then cannot explain what they did.
The argument for under-specifying is straightforward. In a real organisation, nobody hands two teams a complete interface definition. What exists is a partial brief, an assumption on each side, and a date. The teams that succeed are the ones where somebody notices the gap early, walks over, and asks the awkward question about what exactly we mean by ready. Under-specification reproduces that, and it is the only way to reproduce it, because a room given complete instructions can only demonstrate compliance.
Run it as a choice, with three settings.
Named blanks. The brief lists the six headings above and says explicitly that the teams must fill them in. This is the standard setting and the one to use with most groups. Teams cannot fail by never noticing there was a question, which is the failure that feels unfair, and they can still fail by agreeing a number badly, which is the failure that teaches.
Unnamed blanks. The brief states the envelope, the component count and the test, and says nothing about interfaces at all. Teams get the same ten minute conference but nobody tells them what it is for. Some rooms use it to discuss the theme. This produces a much sharper session and a much higher chance of a visible mismatch, and it should only be run with the repair window and second attempt promised in advance.
Conflicting blanks. Advanced and slightly unkind. Each team's paper brief contains one dimension the others do not have, and two of those dimensions are incompatible. This turns the conference into a genuine reconciliation exercise, and it models the real situation where each side is working from a document the other side has never read. Only run this if the debrief is built around it.
The bridge is where under-specification bites hardest, and it is worth choosing the brief for that reason alone. In a city build, an under-specified boundary produces a road that changes width at a district line, which is visible and mildly embarrassing and does not stop anything. In a kingdom build it produces almost nothing at all, because the components are decorative and adjacency is the only requirement. In a bridge build, two halves built to different deck heights simply do not meet. The structure does not exist. The test cannot be run. There is no partial credit available at the physical level, which is why the bridge version produces the conversation that the other briefs only gesture at.
If your group's real problem is that two functions each think the other one owns the handover, run the bridge, name the blanks, and let them find out in ninety minutes what usually takes them a quarter.
Talking to the team next door: three ways to set the window
This is the most important lever on the page. More than the material, more than the scoring, more than the difficulty tier. How much contact you allow between neighbouring teams during the build determines what the session is about, what the finished structure looks like, and whether the debrief has anything in it. Set it deliberately and publish it in the opening brief.
Note first that the lever is only real if the room is arranged for it. Five teams at five tables in one function room will overhear each other regardless of what you announce, and a restriction that everybody can casually breach teaches nothing except that rules here are decorative. If you want a genuine restriction you need physical separation: separate rooms, a screened area per team, or at minimum enough distance and background noise that a normal conversation does not carry. If you cannot achieve that, be honest and define the rule as what it actually is, which is a rule about walking to another team's table rather than a rule about information.
Option one: an open channel
Teams may contact each other at any time, walk anywhere, and look at any component.
What it produces. The highest chance of a structure that fits and the lowest chance of an interesting session. With unrestricted contact, the interface stops being a specification problem and becomes a continuous negotiation, which is easier and less instructive. Two other things happen reliably. The room converges into one large committee, and the pace drops, because every decision now has five stakeholders. And the loudest team's preferences quietly become the standard, not because they argued best but because they walked over most.
When to choose it. A group that has never worked together and where the objective is genuinely just contact between strangers. A very short session where a failed assembly would leave no time to recover. A group with a large proportion of people who find the format difficult and need the reassurance of being able to check.
What you lose. The specification. If teams can check continuously, nobody has to write anything down, and the discipline of holding to an agreed number without supervision is never exercised. That discipline is the thing this format is for.
Option two: scheduled windows, which is the default
Contact is forbidden except in announced blocks. The recommended pattern for a ninety minute build is three windows.
The specification conference, ten to fifteen minutes, before the build starts. All delegates at the taped footprint. Everything the central brief left blank gets settled here, written on the flipchart in numbers, and photographed.
The mid build check, five minutes, at the halfway mark. One delegate per team, at the footprint, with a tape measure. Each states out loud the current actual dimension at their boundary, measured, not remembered. This window is where teams discover that what they built is not what they agreed, at a point where there is still time to fix it.
The pre assembly check, five minutes, fifteen minutes before the freeze. Adjacent pairs only. Last chance to reconcile. After this, nothing.
Why three and not two. The first window is planning, and planning is always optimistic. The second window is the only one that produces measurement, and measurement is the entire point. The third window is repair. Cut any of the three and you lose a distinct function. Organisers who run two windows almost always cut the middle one, which is the one that matters.
Two restrictions worth adding to the windows. Send one nominated delegate per team, not the whole team, which stops the window becoming a plenary and puts a named person on the boundary. And require the delegate to carry the written specification and a tape measure, so the conversation is about numbers rather than about intentions.
What it produces. The best balance available. Teams have enough contact to succeed and not enough to avoid committing to a number. The mismatches that do occur are traceable to a specific moment in a specific window, which makes them excellent debrief material. This is the setting to run unless you have a reason not to.
Option three: one conference, then silence
A single specification conference at the start, then no contact of any kind until assembly. No walking, no messages, no looking.
What it produces. The purest version of the specification lesson and the highest chance of a visible failure. Every ambiguity in the written specification survives untested for ninety minutes and then presents itself in front of the whole room. Teams discover that "the deck sits at about a hand's height" was not a specification, that nobody asked what the datum was, and that two people left the conference with different numbers in their heads and neither checked.
When to choose it. A group whose real problem is that they do not check assumptions with each other, and who need to see the cost rather than be told about it. A group that has run the format before at the scheduled window setting. A senior group that has asked for something hard.
When not to choose it. A first session. A group that has just been through a reorganisation and is already fragile. Any session where you have promised anybody a photograph of a finished structure.
The two things you must do if you run it. Promise both a repair window and a second attempt when you brief the room, so the opening assembly reads as a trial run rather than a verdict. And be ready to debrief a failure properly, which means opening the conversation at the specification rather than at whose component was wrong.
A written-only variant worth knowing about. Contact is allowed but only in writing, on numbered notes handed through a facilitator, with a stated turnaround of five minutes. Slower than speech, forces precision, and produces an artefact you can read out in the debrief. It suits groups whose real communication is asynchronous and written, which is most groups now, and it is a better simulation of their actual working conditions than a face to face conference.
Whatever you pick, publish it before the build starts. A group that assumed it could check with its neighbour and finds out at minute sixty that it cannot will treat that as a trick, and the room will be lost. Announce the rule, announce the windows and their times, and put the times on a visible flipchart.
Choosing a material, and what each does to build time, tolerance and cost
The material decides how precise a component can be, how long it takes to build, whether it will stand up, and what the session costs. It is not a detail to be left to a supplier.
Corrugated cardboard. The default for large structures and the cheapest by a wide margin. Published local retail for a medium corrugated sheet roll of 47 inches by 100 feet is about $22.50, which is enough material for several components, and reclaimed cartons cost nothing at all.
Build time: slow. Cutting corrugated board accurately takes real time and produces a lot of waste. Budget at least half your build window for cutting.
Tolerance: poor and getting worse as the piece gets bigger. A hand cut edge over a metre drifts, corrugated board crushes under a firm hand, and a taped joint adds thickness that nobody accounted for. Assume plus or minus five millimetres on any dimension and design the tolerance clause around that.
Structure: excellent, but only if the teams discover folding and triangulation. Flat cardboard carries almost nothing. Folded into a box section, a triangle or a corrugated stack with the flutes running vertically, it carries a surprising amount. Whether the teams find this out is one of the more satisfying things to watch.
Cost and risk: very low cost, and the one real risk is blades. Cutting corrugated board needs a craft knife, a craft knife needs a cutting mat and a metal rule, and a room of thirty adults with thirty blades and no briefing is a genuine hazard. Either brief it properly and supply mats, or specify pre cut blanks and scissors only.
Foam board. The precision option. Cuts cleanly with a knife, holds a crisp edge, is light, and takes a pin or a pin and glue joint well.
Build time: fast. A clean cut takes seconds and does not need to be trimmed twice.
Tolerance: the best of the sheet materials. Plus or minus one to two millimetres is achievable by an amateur, which means a tight interface specification is actually holdable and the exercise becomes about agreeing the number rather than about hitting it.
Structure: good in compression, poor at the edges, and it dents. Foam board towers stand well and foam board decks sag over anything much beyond half a metre without a rib underneath.
Cost: several times the price of the same area in corrugated board, and local retailers did not publish a per sheet price on the pages read, so price it by quotation. Budget it as the expensive choice and use it where precision is the point.
Wooden blocks and construction planks. Identical modular units, no cutting, no glue, no waste, fully reusable across sessions.
Build time: very fast, and almost all of it is design rather than fabrication, which shifts the balance of the session towards thinking.
Tolerance: effectively perfect within a component, because every unit is the same size, and completely unforgiving between components, because nothing fixes to anything. A gravity structure has no joint at all: two halves either bear on each other or they fall over. This produces the most dramatic assemblies of any material and the highest failure rate.
Structure: real. Load, balance and stability are genuine problems and cannot be taped around, which is exactly why some facilitators prefer it.
Cost: local retail prices for large plank sets were not published on the pages read, and imported sets are the usual source. Treat it as a capital purchase that pays back over several sessions rather than a per session cost.
Interlocking plastic brick systems. The most expensive material by a distance and, for this specific exercise, usually the wrong one.
Build time: fast, familiar and enjoyable.
Tolerance: zero, in the good sense. The system has a fixed grid, so the interface can be specified perfectly in a single sentence: the road leaves your plate eight studs wide, centred, at plate level. Two teams following that will mate exactly.
Cost: high. Bulk plates are published at around $5 for a hundred pieces at a local specialist retailer, and a structure large enough for five teams runs into a serious number.
The reason to think twice: the grid solves the interface problem for you. A modular system with a universal connection standard removes precisely the negotiation the format exists to create. If you have chosen this family because you want teams to argue about a datum, a material with a built in datum is working against you. Choose it when the objective is a polished artefact, a themed showpiece or an accessible build for a mixed group, and choose something cuttable when the objective is interface discipline.
Mixed and supplementary materials worth having regardless. Paper and card tubes for columns, wooden dowels or bamboo skewers for bracing, masking tape in preference to clear tape, since it lifts off surfaces cleanly, string, and a small quantity of hot glue only if you are prepared to supervise it. Ban anything that burns, anything that shatters, anything liquid, and any fixing that makes the structure impossible to take apart, since the components have to be carried to the footprint and adjusted during assembly.
One rule that applies to every material. Issue identical kits, in sealed and labelled boxes, opened at the same moment. The cost of a fairness argument at the end is far higher than the cost of a few extra sheets.
Scale, load, and whether the thing will actually stand up
This is the constraint teams ignore completely until the moment it defeats them, and it is the one genuine piece of engineering in the format.
A small model is forgiving. Scale it up and it stops being forgiving very quickly, because the weight of a structure grows faster than the strength of the parts holding it. A cardboard span that carries a marble at 30 centimetres will sag under its own weight at 120 centimetres, and the team that built the 30 centimetre prototype and scaled the drawing has learnt nothing about what will happen. This is not obvious to a room of adults who have not built anything since school, and it should not be explained to them at the start. It should be built into the test so they meet it.
Put a real load in the specification and state it in the opening brief. A finished object that only has to look like a bridge will be built to look like a bridge. Options that work, in rough order of severity:
A rolling load along the full length. A remote controlled car, a toy vehicle or a ball with a stated weight, which must travel the whole deck without stopping or falling through. This is the best test for a bridge, because it tests every joint in sequence and it produces a genuinely tense ninety seconds.
A static load at the worst point. A stated weight, commonly half a kilogram to two kilograms, placed at the centre of the longest span and held for ten seconds.
A distributed load. A row of identical weights placed along the deck.
A lift test. For a city or a landmark, the finished structure must be liftable or slideable by two people without shedding anything.
Where load actually fails, in practice. Almost never in the middle of a well built component. It fails at the joint, because the joint is the discontinuity: the deck is thinnest there, the fixing is improvised, and it was made in a hurry during assembly by people who did not build either side. A joint that is merely taped is a hinge. A joint with an overlap plate underneath is a beam. That difference is worth a sentence in the specification and it will decide the test.
The three things teams should be told to do and usually are not, which is why you should not tell them:
Triangulate. A rectangle made of card and tape is a mechanism, not a structure, and it will parallelogram sideways. One diagonal makes it rigid. Watching a team discover this at minute fifty is one of the pleasures of running the format.
Put the material on edge. A sheet lying flat carries almost nothing. The same sheet folded into a box, a channel or a trussed rib carries many times more.
Support at the joint, not near it. A pier placed 15 centimetres back from the boundary leaves an unsupported cantilever at exactly the point where the load has to cross.
Set a maximum height and enforce it. Not for the exercise, for safety. A three metre cardboard tower assembled by a room of thirty people will eventually fall on somebody, and the moment somebody stands on a chair to place the top piece you have a real injury risk in a function room with no first aider. State a maximum height that everyone can reach standing on the floor, and state that furniture is not to be stood on.
Sizing the room: people per component, components per structure
The design variable is the component count, and headcount follows from it rather than the other way round.
People per component: four to six, and six is the working maximum. Below four there is nobody free to be a delegate during a communication window without stopping the build. Above six you have the problem this format shares with every vehicle or object build, which is that a single component of a metre or so has room for about two pairs of hands at a time. The rest watch. That is covered in its own section below and it is solved by structure, not by exhortation.
Components per structure: four to eight. Below four the structure has too few boundaries for the interface problem to be the point of the day, and with three components the two joints can be managed by a single conversation. Above eight the assembly gets long, the accumulated tolerance error becomes large enough that even well built components will not close, and the room spends the last half hour watching a small crew struggle. Five is the number that produces the cleanest session and it is the number used in the worked example.
Practical headcount range: 20 to 48 for one structure. Five components of four is twenty. Five of six is thirty, which is the sweet spot. Eight of six is forty eight, which is the practical ceiling for one structure, one footprint and one facilitator team.
Above roughly fifty people, build more than one structure. Two or three parallel structures of five components each, assembled side by side, works far better than one structure with fifteen components. Published operator bands go to five hundred and above, and at that scale you are certainly buying multiple parallel structures, a production crew and a staged finale. That is a legitimate product and it is a different one: parallel structures quietly remove the room wide dependency, because a team's outcome now depends on four other teams rather than on everybody. Ask any supplier quoting for a large group how many separate structures they intend to build, because the answer changes what you are buying.
How to cut the teams. Cut across departments, on purpose, before the day, and publish the list at the door. The value of this format lives at the boundaries, so place the people whose real handovers are broken on opposite sides of a joint. If procurement and engineering do not talk, they should be building spans two and three. That single act of casting does more than any facilitation on the day, and it costs nothing.
One placement rule. Do not put the most senior person in the room in an end component. End components own one joint and carry the least interface risk. Put seniority in the middle, where two neighbours have to be managed at once.
Two clocks and a scheduled assembly, not a leftover ten minutes
Three blocks of time, and organisers routinely plan the first and improvise the other two.
The build clock: sixty to ninety minutes. Below sixty minutes with a cuttable material you get flat, untested components and no time for anything to be discovered. Past two hours the room runs out of problems and teams start decorating, which is fine but is not what you are paying for. Ninety minutes is right for five components of six people in cardboard. Cut it to sixty for modular materials that need no fabrication.
The assembly clock: twenty to thirty minutes, on the run sheet, with a start time. This is the section to read twice, because the single most common structural error in running this format is treating assembly as the thing that happens at the end if there is time. It is not a formality and it is not five minutes. It is a distinct operational phase with its own crew, its own order and its own failure modes.
Why it takes as long as it does:
Components have to be moved. Five components of a metre or more each, carried across a room by people who did not build them, past thirty spectators, without catching a doorframe. That alone is five to ten minutes and it is where components get damaged.
They have to be placed in a stated order. A bridge assembled from both ends inward closes at the middle, which is where you want the accumulated error to appear because it appears once. A bridge assembled left to right accumulates error progressively and the fifth team is punished for four other teams' drift. Decide the order in advance and announce it.
The joints have to be made. Even a simple taped butt joint on five boundaries, done carefully, with someone holding each side, is several minutes. Adaptor pieces, if you have specified them, are cut and fitted here.
Things go wrong here and need time to be fixed. A joint that does not close needs a decision, and the decision needs minutes that exist.
How to staff it. Name an assembly crew of two or three people in advance, one of whom should be the specification keeper who has carried the tape measure all session. Everyone else stands behind a taped line. An assembly where thirty people all place their own component takes twenty five minutes, contains four errors, and has no moment in it.
The repair window: ten to fifteen minutes, promised in the brief. After the first test. It is the block in which the most interesting behaviour of the day shows up, and where you find out whether the teams whose joints closed go and help the team whose joint did not.
A realistic full session. Fifteen minutes brief and specification issue, fifteen minutes specification conference, ninety minutes build with two five minute windows inside it, five minutes freeze and stage, twenty five minutes assembly, five minutes test, twelve minutes repair, five minutes second test, ten minutes presentations, fifteen minutes debrief. Add that up and it is three hours and thirty seven minutes with nothing spare, so publish four hours and put a ten minute break at the freeze. The published one hour listings are describing a build window, not a session.
The kit list, the floor area and the surface underneath
Per team, for a cardboard or foam board build:
Sheet material, identical quantity per team, in a sealed labelled box
Craft knives with retracting blades, or scissors only if you have chosen not to brief blade safety, plus one cutting mat per two people
A metal rule or straight edge, which doubles as a cutting guide
One tape measure per team, which stays with the specification keeper
Masking tape, one or two rolls per team, and a little stronger tape held back for use at the joints
Pencils and markers, and a marker that writes on the material
String, dowels or skewers for bracing
Paper and a pen for the written specification, printed and issued
Shared, for the room:
The taped assembly footprint on the floor, marked and numbered
A flipchart for the agreed specification, kept visible all session
A second tape measure and a spirit level for the specification audit
The test load, prepared and weighed in advance
A camera on a fixed position for the test, because the footage is the only reliable record of where it failed
A bin and a broom, because cardboard produces an astonishing amount of offcut
Floor and space. This format needs more floor than almost anything else at the same headcount, and the requirement is rarely in a quotation. You need a build table or floor area per team, roughly two to three square metres each, plus a clear assembly footprint large enough for the finished structure with a metre of circulation all round, plus room for thirty people to stand behind a line and watch. For thirty people and a six metre bridge that is a genuinely large room. The published planning allowance for a seated function is 30 to 40 square feet for each guest, and this format wants more, because the assembly footprint is dead space for the first two hours and then becomes the whole event.
The surface matters. A flat, level, hard floor is what you want, and a carpeted function room floor is acceptable. What causes problems is a floor with a slope, which some multi purpose halls have for drainage, and a floor with a thick soft carpet or an expansion joint running through where the structure has to sit. Walk the room and check. If the assembly is on tables rather than the floor, the tables must be the same height as each other and locked, which is worth checking with a straight edge before anybody arrives.
Venue cost. Rates published for hiring a function room in 2026 span roughly $40 up to $250 for each hour, and a three hour minimum is close to universal. Your own office is very often the better answer if it has the floor area, and the material budget for a thirty person cardboard build is small enough that an in house session is realistic for most organisations.
Marking the joints rather than the towers
The scoring is where organisers accidentally undo everything else on this page. Score the components and you have told five teams to optimise the thing they control, which is exactly the behaviour that breaks the structure. Score the joints and you have made the boundary worth caring about.
The argument, stated plainly
A component is built by one team, judged as one team's work, and improved by that team working alone. A joint is built by nobody, owned by two teams, and improved only by two teams talking. If the scoring is silent about joints, then rationally every team should spend every minute on its own component, and rationally every team will. The joint then belongs to nobody, which is the defining failure of this format and the one you should design against rather than lament afterwards.
There is a second reason, which is that joints are objectively measurable and components are not. Whether a joint closed is a fact. Whether a tower is impressive is a taste. A scoring scheme dominated by facts survives contact with a disappointed room in a way that a scheme dominated by judgement does not.
The recommended scheme
A hundred points per team.
Your own component stands unaided and carries the test load across its own length: 20 points. All or nothing, checked before assembly.
Joint integrity: 30 points, divided by the number of joints your team owns, and awarded to both adjoining teams or to neither. A team in the middle of a bridge owns two joints at 15 points each. A team at the end owns one joint worth the full 30. Normalising this way keeps the maximum equal for every team while concentrating an end team's entire joint risk in one place, which is a real and interesting difference the room will notice. Award or withhold each joint to both teams together. This clause is the engine of the whole scheme. A boundary owned by two teams now carries a score owned by two teams, and no team can defend its own total by leaving the boundary to somebody else.
The whole structure passes the test: 25 points, to every team or to no team. A collective term that no team can win alone and no team can lose alone.
Specification compliance audit: 10 points, measured, not judged. After the freeze and before assembly, walk the components with the flipchart and a tape measure and check each one against the agreed numbers. Two points per dimension checked, awarded or not. Nobody argues with a tape measure, and the audit is the moment a team finds out their component is not what they believe it is.
Design and build quality: 15 points, judged. Ingenuity, use of material, structural cleverness, finish, and whatever detail earns a laugh from the room. A panel judges it, against criteria announced before the build.
Interface and collective terms are 85 per cent of the total and judgement is 15. Put those weights in the opening brief. Weights nobody knows about cannot change how anybody builds. A spectacular component whose joints both fail scores at most 45. A plain component that stands, closes both joints and is part of a working structure scores 75 before a judge has looked at it.
Which attempt counts. Take the best one. Marking the opening test punishes teams for something they had no way of knowing and empties the repair window of any purpose, and the repair window is where the behaviour you are after actually shows up. If the first test needs a stake, attach a five point bonus to passing it and stop there.
Two variants worth having
The single room score, for a group where competition is the problem. Drop all individual terms. The room gets one number: the structure passes and everyone scores a hundred, or it fails and the score is the number of joints that closed as a percentage of the total. Nobody is ranked. This suits merged departments and post reorganisation groups where a leaderboard would work directly against the reason you booked the day. You lose the individual accountability that the component term creates, so use it knowingly.
The bidding variant, for a group that will not take anything seriously without a stake.Before the freeze, each team writes down whether they believe each of their joints will close. A correct prediction is worth five points, an incorrect one costs five. It is cheap, it takes two minutes, and it forces every team to convert a vague feeling about their boundary into a stated position. The teams that write down "no" and then go and do something about it in the last fifteen minutes are the ones to watch.
Roles, and the two people who end up doing all the building
This format has the same problem every physical build format has, and it is worth stating without euphemism. A component roughly a metre long has room for about two pairs of hands. In a team of six, two people build and four people watch, and the four who watch are usually the same four every time: the less confident, the less physically comfortable, and the ones who did not get to the material box first. Ninety minutes of watching is worse than not attending.
It is a structural problem and it has a structural fix. Do not solve it by asking people to include each other.
Split the component into sub assemblies from the first minute. A bridge span is not one object. It is a deck, two piers, a bracing system, a joining face at each end and a deck surface. Those are four or five workstreams that can be fabricated in parallel at different corners of the same table and brought together at minute sixty. Say this in the brief, and say it as a requirement: your component must be built as at least three parts made simultaneously. That one instruction converts two working pairs of hands into six.
Name five roles and require every team to fill them. Name them in the invitation so people can choose rather than being assigned.
The build lead, who is accountable for everything inside the component and settles internal arguments.
The specification keeper, who holds the written specification and the tape measure, checks the component against the agreed numbers at intervals, and holds the sole authority to declare something out of specification. This is a genuinely load bearing job for somebody who does not want to cut cardboard, and it should be presented that way.
The joint delegate, one per boundary, named at the start, who attends the communication windows and owns the interface with a specific neighbour. A middle team needs two of these. Putting a name against every boundary before the build starts does more good than any other single decision an organiser can make, which is precisely why you might hold it back on a day designed to let a room find out for itself what an unowned boundary costs.
The materials and cutting lead, who manages the kit, does the measuring and marking before anybody cuts, and keeps the offcuts usable.
The presenter, who watches the build rather than doing it and has ninety seconds to explain the component at the end.
Rotate the tools at the halfway mark, out loud. Announce it as a rule rather than leaving it to the teams, because the person holding the knife at minute forty five will still be holding it at minute ninety unless somebody makes them stop.
Two things to avoid. Never let the room pick its own teams: given the choice, the confident cluster together and everybody else is left over. And never concentrate the reluctant in a single group, which is what letting people choose usually produces by accident.
Thirty people, five teams, one bridge: the session in full
A worked example, with the arithmetic, showing how the team that builds the best component can finish last.
The setup. Thirty people, five teams of six, one function room with a clear floor eight metres long. Material is corrugated cardboard, masking tape, dowels, one craft knife and cutting mat per two people, one tape measure per team. Session length four hours including a break.
The central specification, issued on paper:
The finished object is a bridge with a total span of 6.00 metres, spanning a taped footprint on the floor.
Five spans, numbered one to five from left to right, 1.20 metres each. Team numbers match span numbers.
Every span must be self supporting. Nothing may rest on the floor between the marked pier zones.
The test: a toy vehicle weighing 800 grams is placed at the left abutment and must travel the full 6.00 metres and off the right abutment without stopping, tipping or falling through.
Materials, quantities and the clock.
Maximum height 900 millimetres. Nobody stands on furniture.
What the specification deliberately does not say: deck height above the floor, deck width, the datum from which any height is measured, how two spans are joined, the tolerance at a joint, or anything at all about pier design.
The communication windows, published in the brief: a fifteen minute specification conference before the build, one delegate per team at the footprint. A five minute measured check at the forty five minute mark. A five minute pre assembly check at the seventy five minute mark, adjacent pairs only. No other contact.
What happens in the conference. Five delegates stand at the taped footprint for fifteen minutes. They agree, and write on the flipchart, that the deck will be 200 millimetres wide, that spans will butt against each other and be taped, and that the deck will sit at "about a hand's height" above the floor. Nobody asks what a hand's height is in millimetres. Nobody fixes a datum. Two delegates leave believing the number is roughly 150 millimetres and one leaves thinking it is more like 200. The conference finishes four minutes early and everybody is pleased with how efficient it was.
What happens in the build. Team three has an ex architecture student. They build a genuinely elegant trussed span with a triangulated deck, a curved profile and a deck sitting at 220 millimetres, which their design needed for the truss depth. It is the best object in the room by a distance. Teams one, two, four and five build plain box girders with the decks at somewhere between 145 and 160 millimetres.
What happens at the forty five minute window. This is the window that would have saved the session. Team three's delegate goes, but takes no tape measure and describes the span rather than measuring it. The word used is "high" and everybody nods, because at that point every span is a pile of unassembled parts and "high" carries no information. Team two's delegate, who has a tape measure, states 150 millimetres out loud. Nobody connects the two statements.
What happens at the seventy five minute window. Team three's delegate and team two's delegate stand next to their two half built spans and one of them says the decks look different. There are fifteen minutes left. Team three decides that cutting the piers down would destroy the truss geometry and that the joint can be adapted with a ramp at assembly. Nobody tells teams four or five.
The compliance audit. With the tape measure, against the flipchart: deck width is within tolerance on all five spans, so every team takes those points. Deck height was never written as a number, so it cannot be audited, which is itself the finding, and the facilitator says so out loud.
The assembly. Named crew of three, from both ends inward, twenty five minutes. Joints one to two and four to five close cleanly. Joint two to three has a 70 millimetre step. Joint three to four has a 65 millimetre step. The improvised cardboard ramps are made during assembly, and they hinge.
The test. The vehicle crosses joint one to two, climbs the first ramp, reaches the top of span three, descends the second ramp and stops with a wheel through the ramp. The bridge does not pass.
The scores, first attempt:
TEAM | OWN SPAN | JOINTS | STRUCTURE | AUDIT | DESIGN | TOTAL |
|---|---|---|---|---|---|---|
1 (end, 1 joint at 30) | 20 | 30 | 0 | 8 | 9 | 67 |
2 (middle, 2 at 15) | 20 | 15 | 0 | 8 | 11 | 54 |
3 (middle, 2 at 15) | 20 | 0 | 0 | 8 | 15 | 43 |
4 (middle, 2 at 15) | 20 | 15 | 0 | 8 | 10 | 53 |
5 (end, 1 joint at 30) | 20 | 30 | 0 | 8 | 12 | 70 |
Team three built the best span in the room, took every available design point, and finished last. Not because they built badly. Because they solved their own component beautifully and treated the boundary as somebody else's problem, and the scoring made that cost visible. Teams two and four, who did nothing wrong, each lost fifteen points to a neighbour's decision, which is the second uncomfortable lesson of the day and the one that transfers most directly to a real organisation.
The repair window, twelve minutes. Announced in the opening brief, so nobody experiences the first test as the verdict. What happens next is the thing worth watching. Teams one and five, whose joints closed and who have nothing to gain individually, walk over to span three, and eleven people cut 70 millimetres off team three's piers, rebuild the truss ends, and reset the span. The elegance survives. The bridge is level.
The second test. The vehicle crosses all six metres in nineteen seconds. Everybody takes the 25 collective points, and the recalculated totals are 92, 79, 68, 78 and 95. Team three still finishes last, and finishes last with a working bridge and a room that helped them, which is a substantially better ending than the alternative.
What to say in the debrief, in one line, before the questions start. The failure was not team three's span. It was a phrase in the specification conference that four people heard as a number and one person heard as a description, and nobody measured it until it was too late to be cheap.
Three difficulty tiers, and the single rule that separates them
Change one setting per session. Every item below is a genuine control.
Accessible. Four components. The central brief fixes the mating dimensions and the datum, leaving only the connection method and the internal design to the teams. Open communication channel. Modular material with a built in grid, or pre cut blanks. Ninety minute build. A static load test rather than a rolling one. Two attempts. The structure will stand and the session is about the build.
Standard. Five components. The six specification headings are named in the brief and the teams fill them in during a timed conference. Three scheduled communication windows, one delegate each. Cuttable material, cut by the teams. Ninety minute build. A rolling load across the full length. A repair window and a second attempt. Almost everything on this page is written at this setting.
Hard. Six to eight components. The brief states the envelope, the component count and the test, and says nothing about interfaces. One conference then silence. Cardboard, cut from flat stock. Seventy five minute build. A rolling load plus a static load at the centre. Each team's paper brief carries one dimension the others do not have. Two attempts, with the arithmetic of the situation stated honestly to the room before the build so that a failed first test reads as physics rather than as anybody's fault.
Three settings that resemble difficulty and are not. Shrinking the components, which only makes the cutting fiddly. Piling on extra material types, which stretches the design phase without introducing one new decision. And squeezing the build clock, which yields unfinished components rather than quicker ones, and deletes exactly the time in which the interface problem would have surfaced.
Two further settings worth having in reserve. The blind component, where one team may not see the assembly footprint at any point and works only from the written specification, which is a strong specification discipline lesson and better than making everything harder. And the late change, where at the sixty minute mark you announce that the deck must now carry a load at twice the stated weight, which is cruel, accurate to life, and suitable only for a day whose stated subject is coping with change.
Failure modes specific to a build that must fit together
The datum nobody fixed. Symptom: two components built to the same agreed number, at different actual heights, because one team measured from the table and the other from the floor. No other mistake sinks more sessions in this family, and none is easier to design out. Intervention: make the datum the first item in the specification conference, state it out loud, write it on the flipchart, and give one tape measure per team to a named specification keeper who is the only person allowed to declare a dimension.
The agreement made in words. Symptom: "about a hand's height", "roughly the same as yours", "shoulder width". Every one of these is an agreement that two people leave with different numbers in their heads. Intervention: a stated rule that no dimension is agreed until it is written on the flipchart in millimetres, and a facilitator who walks past the conference once and asks what the number is.
Two people building and four watching. Symptom: at minute forty, two people are cutting and taping and four are on their phones. Intervention: the sub assembly requirement in the brief, the five named roles, and an announced tool rotation at the halfway mark. If it is already happening, walk over and ask the build lead which three parts are being made in parallel right now, and wait for the answer.
Assembly discovered as an afterthought. Symptom: the freeze is called, and it becomes apparent that nobody has decided who carries what, in what order, through which doorway, onto a footprint currently occupied by two tables. Intervention: put assembly on the run sheet with a start time, name the crew in the opening brief, mark the footprint before anyone arrives, and rehearse the carry with empty hands during the break.
The structure that will not stand. Symptom: everything fits, the last joint closes, and the whole thing sags or folds when the load is applied. Intervention: put a real load in the specification at the start so teams build for it, insist that every component is load tested by its own team before the freeze, and pay attention to the joints during assembly, because the joint is where the structure is thinnest and it is the part built last by the fewest people.
The team that solves its component by taking its neighbour's space. Symptom: a span that is 1.28 metres long, or a district that has quietly annexed four centimetres of the boundary road. Malice is almost never involved. A team has simply solved its own problem by borrowing space at the edge. Intervention: the marked component boundary on the taped footprint, an audit with a tape measure before assembly, and a facilitator who walks the tables at minute sixty.
The open channel that becomes a committee. Symptom: at minute fifty, eleven people are standing at the footprint having a design conversation and nobody is building. Intervention: scheduled windows with published times, one delegate per team, and a facilitator who ends the window on the clock.
Blades, glue and height. Symptom: a cut finger, a burn, or somebody on a chair. Intervention: a genuine ninety second safety brief, one cutting mat per two people, a rule that cutting happens only on a mat with a metal rule, hot glue only under supervision or not at all, a stated maximum height, and a first aid kit in the room.
Interface negotiation is the output, and speed is not
What this genuinely builds.
Interface negotiation, which is the primary output and is rare. The mechanism is specific and it is worth stating precisely, because it is the reason to choose this format over anything else in the catalogue. Five teams have to write a specification for a boundary before any of them knows what they are building, hold to it for ninety minutes without supervision, and then find out in public whether they were describing the same object. There is no way to fudge the finding. Very few exercises produce a result that cannot be argued with, and this one produces it on a table in front of thirty people.
The discipline of checking an assumption against a neighbour. This is the transferable habit and it is narrower and more useful than "communication". The specific behaviour is: I believe the deck is at 150 millimetres, I am not going to assume that you also believe that, I am going to walk over with a tape measure and find out. Teams that do this at minute forty five succeed. Teams that do it at minute seventy five spend the repair window fixing it. Teams that never do it discover the answer during the test. The format compresses that entire arc into an afternoon, and it gives a group a phrase they can use in a real meeting afterwards, which is more than most days out achieve.
A written specification habit. A group that has watched a session fail on the word "about" will write numbers down for a while afterwards. That is a small, concrete and genuinely durable outcome.
A shared physical object. Like the canvas format, this one produces something that exists after the day ends. Unlike the canvas, it usually cannot be kept, because a six metre cardboard bridge does not fit anywhere. Photograph it properly, straight on, before anyone touches it.
What it does not build, and should not be sold as building.
Not speed, and not pressure. These sessions run long and they run calm. There is no clock that punishes hesitation minute by minute, no incomplete information forcing an immediate bet, and long stretches where the room is quiet and people are measuring things. If the brief was energy, competition and a group that needs to sweat, this is the wrong format and it will read as a disappointment no matter how well you run it.
Not an icebreaker. It is slow, technical and detailed, and in the first twenty minutes the confident and the practical will take over while everybody else works out where they fit. For a room of strangers, put something loud earlier in the day and this after it.
Not decision making under uncertainty. The problem is tightly specified, every constraint is visible and the failure conditions are known in advance. What is tested is execution, coordination and the willingness to check, not judgement under ambiguity.
Not conflict resolution. Two teams that are not talking will build two components in silence and discover at assembly that they do not fit, which is an accurate diagnosis and not a treatment. The format shows a group that they have an interface problem. It does not fix the relationship underneath it.
Not equally accessible without design. Cutting, kneeling, reaching, fine measurement and long spells on your feet are all genuinely required. Collect that information on the sign up form and never by a show of hands, and make the specification keeper and presenter roles visible and real, so that taking one reads as a choice rather than as a retreat.
What to ask once the structure is standing
Fifteen minutes, seated, in the room, immediately after the final test and before anyone packs up. This is where the afternoon becomes worth its cost, and hardly any supplier builds it into a quotation.
Two rules before you start. Do not open by asking how everyone found it. And call first on a team whose joint failed to close, so that the conversation starts at a boundary instead of at whose component looked best.
About the specification:
What did the conference agree that turned out not to be a number? Read the flipchart out loud and find it.
What did you assume everyone else understood, without checking, and what made you confident enough not to check?
Which of the six headings did nobody raise at all, and why do you think that one got missed?
Run it again tomorrow: which single clause would you insist on before anybody picked up a knife?
About the joints:
Who owned the joint between spans three and four? Name that person now, out loud.
For the boundary you owned, at what point did you know there was a problem, and what did you do in the next five minutes?
Which of you went over to another team unprompted today, and what was the errand?
Did any boundary sit there with each side assuming the other had it in hand? For how long, before somebody noticed?
About the communication windows:
In the forty five minute window, what did you actually say, and did you bring a measurement or a description?
What did you find out in a window that you could not have found out on your own?
Was there something you knew at minute fifty that you did not tell your neighbour? What stopped you?
If you had been given one more window, when would you have wanted it and what would you have used it for?
About the trade off inside a team:
At what point did your team stop improving the component and start protecting the interface? What triggered the switch?
Did anybody argue for a change that made your own component worse and the structure better? What happened to that argument?
About what happened after the first test:
When the first test failed, what did the teams whose joints had closed do in the following sixty seconds?
The questions that carry it back to work, which is the whole reason for the afternoon:
Name a boundary in our actual work that each of two teams quietly expects the other side to own.
Which of our handovers has been specified in words rather than in numbers, and who would have to agree the numbers?
What is our version of the forty five minute measured check, and why do we not currently do it?
Finish on question sixteen and then say nothing for a while. Once a group can point at a real boundary inside their own organisation, a room full of cardboard has become an action with somebody's name on it. Write the answer up on the flipchart, take a photograph of it, and circulate that the next morning.
Generic labels for the format, and how to interrogate a proposal
The generic terms in use for this family are collaborative construction, collaborative build, modular build, sectional build, mega build, giant build challenge, group construction challenge, structure challenge, bridge build and city build. Any of those will find the right kind of supplier in a search.
Beyond the generic terms, the market behaves the way it behaves everywhere else in this catalogue: one mechanic circulates under a wide spread of invented product names, most coined in house by whoever is selling, so the same underlying activity turns up wearing a badge that exists nowhere else. Counting the badges tells you nothing about which proposal is better. Judge the mechanic.
There is one real difference buried in the labelling, and it is worth money. Some products in this family assemble the components into one object and test it. Others have each team build an independent model on a shared theme and then place them next to each other for a photograph. The second one is a different and much easier activity: it has no interface, no fit, no tolerance and no failure condition, and it should not cost what the first one costs. The word "combined" in a proposal covers both. Ask which.
Nine questions that work regardless of what the product is called:
How many components, and how many people per component?
Are the components physically joined into one structure at the end, or are they judged separately and photographed together? If it is the second, this is a different activity.
Is there a load test or a functional test on the assembled structure, and what exactly is it?
Who writes the interface specification, the facilitator or the teams?
If the teams write it, how many minutes are scheduled for that, and is it before or after the build starts?
May teams talk to their neighbours during the build, when, and for how long?
How many minutes are scheduled for assembly, as distinct from the build?
What happens if the components do not fit? Is there a repair window and a second attempt?
What material, how much floor area do you need, and are you bringing tables?
Questions two, four, six and seven will tell you whether a supplier has designed this format or merely shelved it.
Where it parts company with its nearest neighbours.
The chain reaction build lines up segments that trigger each other and runs on one release. Failure there is sequential and total, and what crosses a boundary is energy rather than matter, so two segments never have to share a dimension. Here the components must physically fit, and the failure is a step you can measure with a ruler.
The collaborative canvas splits an image into painted panels. The split is visual, there is no physical join, nothing bears on anything, and a mismatched edge can be absorbed as style. Here a mismatch cannot be hidden from anybody in the room.
An independent tower or structure build, where each team builds its own object from an identical kit and the tallest or strongest wins, shares the materials and none of the mechanism. Nothing combines, so no team depends on any other team, and the interface problem that justifies this format's length does not exist.
A station rotation format runs scored challenges in parallel, with teams moving between them and no team ever handling another team's work. Picking and weighting the stations is the design job there. Here nobody moves at all.
A route based race is about travel, sequence and clue design. This one happens in a single room, standing at a table.
Answers to what HR asks about joined structure builds
What is a collaborative construction team building activity?
A build challenge where every team makes one component of a single large structure, after which the components are brought together into one object that has to physically fit and then pass a test. Bridge and city briefs are the usual ones.
How is it different from a chain reaction build?
A chain reaction passes energy along a line of segments and fails sequentially when one join does not transmit. This format passes nothing: the components must physically fit, share real dimensions, and bear load. A mismatch here is a measurable step that nobody can hide.
How many people does a collaborative construction need?
Twenty to forty eight people for a single structure, split into four to eight components of four to six people each. Five components of six is the ideal shape. Published operator bands run from eight participants to five hundred and above.
How long does a session take?
Four hours is realistic for a full session, covering a brief, a specification conference, a ninety minute build, a scheduled assembly, a test, a repair window, a second test and a debrief. Published listings quoting one hour are describing the build window only.
What does the interface specification need to cover?
Six things: the datum that all measurements are taken from, the mating dimensions in millimetres, the joining face and its orientation, the connection method, the tolerance and who absorbs it, and the load the joint has to carry. The datum is most often forgotten.
Should teams be allowed to talk to each other during the build?
Yes, in three scheduled windows: a specification conference before the build, a measured check at the halfway mark, and a pre assembly check near the end. One nominated delegate per team, carrying a tape measure. Open contact turns the room into a committee.
What happens if you ban contact between teams entirely?
Every ambiguity in the written specification survives untested and appears in public at assembly. That is the sharpest version of the lesson and the highest chance of failure. Only run it with a promised repair window, a second attempt and a careful debrief.
Which material should we use?
Corrugated cardboard for the cheapest version, foam board when precision matters, wooden planks for a fast gravity build with no fixings, and interlocking plastic bricks only when a polished artefact matters more than negotiation, since a fixed grid solves the interface problem for you.
How much does a collaborative construction cost in Singapore?
Nobody local publishes a per person rate for it, so use the general bands: $25 to $40 a head in your office, then $45 to $90 for a facilitated half day in a hired venue, and $100 to $180 at the premium full day tier.
How should it be scored?
Score the joints, not the components. Award 20 points for your own component standing, 30 divided across the joints you own and awarded to both adjoining teams, 25 to everyone if the whole structure passes, 10 for a measured audit and 15 for judged design.
Why score the joints rather than the components?
Because a component belongs to one team and improves when that team works alone, while a joint belongs to two teams and improves only when they talk. Scoring silent about joints tells every team to optimise alone, which breaks the structure.
Why does assembly need its own slot on the run sheet?
Because it takes twenty to thirty minutes, not five. Components must be carried across a room, placed in a stated order, joined by hand, and any mismatch discovered needs minutes to repair. Assembly left to the last ten minutes fails in public.
What does this format actually develop?
Interface negotiation and the discipline of checking an assumption against a neighbour before it is too late to be cheap. It also produces a written specification habit and a shared physical result. It does not develop speed, pressure tolerance or decision making under uncertainty.
Is it a good icebreaker?
No. These sessions run long and calm, the first twenty minutes are dominated by the confident and the practical, and the work is technical and detailed. Run something loud and physical first if the group has never met, and use this second.
Why is the bridge brief harder than a city or a kingdom?
Because two halves built to different deck heights simply do not meet, and there is no partial credit at the physical level. A city absorbs a district a few centimetres out, and a kingdom absorbs almost anything. Take the forgiveness away.
Can you run this at PLAYON?
No. PLAYON has no build space, no materials and no facilitated construction product, so book this one with a corporate build supplier or run it yourself. What PLAYON is, instead, is an indoor venue for active station based formats.