Let them eat cake: engineering risk & responsibility in repair

My name is Kate – I’m a civil engineer by education and an M&E engineer by profession. Right now I work for a start-up – the closest I’ve come to testing some of my hypotheses about repair. I'm supporting two colleagues to build Co-Risk Labs – a worker cooperative – where we want to do more work on what we call Critical Technical Practice (after Phil Agre’s paper on engineers in working in artificial intelligence). This practice element is what I want to touch on today.

So, before I get into the critical part of this talk, I want to say that I’ve always thought of us – people who chose to be trained as engineers – as essentially non-evil individuals.

But I’m come round to a different view after questioning repair decisions here and elsewhere: I think now that being a non-evil individual may be a necessary but insufficient condition for “doing no harm”.


I only started to ponder this in earnest after working for not-for-profit international humanitarian organisations in Haiti.

The trouble is when these experiences crop up in conversation, people have already assumed that if you did aid work you did something good and brave and voluntary. But, in reality, I felt like I’d participated in something shameful.

I came away convinced that bad decisions had been made (by bad I mean both the process of the decision-making and the decisions themselves which were expensive, potentially damaging and arguably self-serving). I started to think about these decisions as taking place in a space between accident and conspiracy: we worked with intention so what came about was not an accident but we were not deliberately trying to ride roughshod over people either. 

In particular, I was interested in why we neglected repair. When you look back at the academic literature, for 40 years, observers of the humanitarian sector have bemoaned:

·      a focus on the product of shelter over and above people’s own processes of sheltering and

·      the sector’s tendency to favour imported “boxed” shelter but

·      the neglect local practices of building and repairing private dwellings.

I felt that in the context of an industrialised aid response arriving in Port-au-Prince, these tendencies had played out again and had been exacerbated by engineers giving advice - at the highest levels - that was structurally sound but culturally inappropriate.

What I want to talk about today are some of the underlying things that were influencing that space between accident and conspiracy. Including the things we take pride in – operating in a rational, systematic, process-oriented and professional manner.

Some of you might have heard me talk about Port-au-Prince before – and I will have to touch on the specifics because it was so formative and extreme for me.

But what I really want to do is open this out into a broader discussion, so I will to start my favourite questions about

·      why money bypasses repair,

·      sketch out some concepts and then

·      look at stories form different settings before bringing us back to contemporary engineering narratives




The money

Let’s start with the money. For me, this is the most interesting and important question: what is it that drives money/capital to flow to particular places and into certain activities? And what can my tribe – engineers - do about this?

From this question flow others: What does this mean for the way we – humans of engineering - make cities? What does it mean for the way people respond to and recover from disasters? How are we – professional engineers - implicated in these decisions? How will technology fossilize or radicalise these flows of money?

In the middle of 2012, I went back to Haiti to evaluate a number of shelter projects funded by one of the major donors. In 2010 following the earthquake, the donor had pledged around $500m (from the figures I can find, stands at about 25% of Haiti’s national budget in 2009).

Of this around $200m had been allocated to the housing sector. (Incidentally, at a national level the very first estimates in the PDNA anticipated about 30% of aid should go to repair but it wasn’t until February 2011 that these even started). And by mid-2012, 80%-90% of that allocation was ear-marked and being spent on transitional shelter kits: based on a timber frame, designed on paper, imported materials and assembly via varying degrees of pre-fabrication and on-site construction. In terms of unit costs and overheads, these varied from:

·      $3,438 per unit to 6,348 per unit excl. overheads (UN Habitat estimated that it cost around $3,500 to build a typical house – confined masonry - from scratch and the benchmark for household incomes in 2009 was about $2,500-3,000 to $12,000 p.a. for the poorest slum inhabitants…

·      Overheads (gross per unit project cost minus material, labour and vehicles) ranged from 15% to 49% and, from our analysis, did not bear much relationship to build quality or economies of scale (500 to 2500). Field based construction supervisors seemed to make the biggest difference to build quality which should make you ask questions about a) what really defines good design b) which factors and which point in the supply chain and procurement systems make a difference to “riskiness of a house” and c) timber was super scarce and had been banned from the city since the 1920s because of fire risks would make supervision of woodworking skills relevant later… OR you could counter and say well if construction is so poor, building anything more complicated than a garden shed is extremely risky.

There isn’t actually a clear answer to whether the kits are right or wrong. But in the aftermath this is almost the only question that gets attention because the kits drew so much money and were the only solution that we repeatedly documented and photographed and debated…

My preoccupation is rather that, at a strategic level, engineers played a part in this decision that was neither process-oriented, systematic or rational and, I will argue, was collectively unprofessional…

a) not systematic - that kits physical form was a reflection of the domination of technical voices and a reflex to both control risk and focus on a single hazard (BUT I’m not convinced this did control risk – and we’ll look at the  shelter sector’s own alogroithm for triage and risk minimization in a minute)

b) not process-oriented – did not take a systemic view - because as the kits took – to the exclusion of all other options – the lion’s share of the money, this “homogenized” the response in the varied, complex texture of the city (BUT of course in the end as the cost and quality and pace of delivery unfolded, it was not at all homogenous and had none of the advantages of homogenous mass production)

c) not rational, the technical response to the risk of earthquakes was to impose high safety standards: this meant spending money on thousands of highly engineered, imported timber shelters and a handful of houses built to Californian codes. In my view, this was like doing heart surgery for a select few when we needed the structural equivalent of soap and water.

The advice

So let’s look at some of the things professionals said (engineers and NGO staff) at the time:

“best thing to do? Knock it down and start again”

Fag break huddle, MTPTC, Haiti

“we trained people but they are still doing just what they did before”

Student recruited to monitor construction

“strategically dismantle the top floor on many of these houses…”

Seismic Engineer’s Assignment Report


“Any organization that takes on a repair program and decides it does not need to meet code, whether national or provisional or borrowed from an advanced country, takes on huge risk, puts the beneficiaries at huge risk, and sets an incredibly poor example in a place that desperately needs to be inspired by good examples from countries that have their act together. We couldn't do this in London...why should it be "good enough" for Port-au-Prince?”


“This isn’t about the earthquake in Haiti, this is about business development for our organization”

[This last one isn’t about earthquakes BUT it is the reason why we have to look at the underpinning optimisations in the algorithms]


The algorithms

Those are some of things we said. What about the processes – the procedures, the routines we used to understand what was happening?

Concepts of Shelter

First is an idea about shelter as a process not a product – if you like, this is one conception of a household’s algorithm, it charts iterative steps – incremental upgrades to their shelter object.

•       Shelter as a process (Davis, 1978)…

•       Housing as a verb (Turner, 1973)…


Repair Algorithm

This comes from one of the handbooks. It looks disarmingly simple and convincing: neat categories, it appears to be optimizing for structural safety and leads to a clear set of interventions. It is appears to be discernible…

It goes like this: is the location deemed safe enough? Ok what about the building? If it’s beyond repair, rebuild it. If it’s not damaged, improve it to meet code. If it’s damaged… estimate the cost of the damage, diagnose the defects that resulted in damage and then “orchestrate” repair…. That is to say: mass-customised, seismically sound repair of masonry structures that are private assets, often not owner-occupied and spread across a massive city).

But let’s look again at this algorithm.

First of all: are the decisions technical assessements of safety, structural diagnoses or feasibility assessments for triage? The real optimization is about whether the cost of starting again is less than the cost of fixing the house given the seismic risk context. But are all the costs in this algorithm?

If my house is damaged:

·      What about the costs and risks of the meantime? (i.e. waiting for a government plan to be announced or an NGO to arrive and save you, waiting for remittances to arrive from relatives in Miami, waiting for builders and materials to be available?)

·      What about my costs and risks of living under canvas?

·      What about the slow, inaccurate engineering that is inevitable given the practical difficulty of mass, bespoke technical diagnosis?

·      What about the costs of my dream home – the one I know have a chance to realise?

Meanwhile, as if this were not complicated enough, the other algorithm that NGOs are deploying is for beneficiary selection – this is optimized for targeting resources to the most vulnerable people. This algorithm goes like this: is this family displaced? Prioritise displaced families. Is this family vulnerable ie. Is it headed by a single female, contain children under 5, elders, chronically sick or disabled? Prioritise vulnerable. Annoyingly, the most vulnerable people do not necessarily live in the most vulnerable buildings…. Targeting collapsed buildings may not mean targeting most vulnerable households… And in fact if your dominant technical solution is the shed, it can only ultimately be delivered to a plot of land – cleared of a destroyed building and can only be received by a household with permission to access the land - urban vulnerability is often characterized by an inability to access land.

So what would you have done? Well, clearly spend money on controllable timber boxes – glorified garden sheds.

And slowly produce some (disputed) technical guidelines for more typical building types.

“efforts to solve the problem only with the dissemination of technical reports for engineers may best be described as a “Marie Antoinette approach to Earthquake Hazard Mitigation” from the quote “…then, let them eat cake”

                                                (Langenbach et al., 2006) Langenbach, R., Mosalam, K.M., Akarusu, S., Dusi, A., 2006



What about the rationale?

So…. These complexities made me think that the way the money was spent was neither accident nor conspiracy (remember I was in the room for those decisions) but it was rooted in our ideas about professionalism and our ways of thinking. Weird things happened. For example, we kept reporting on how well we did what we said we would do BUT we never questioned whether what we said we would do was still the best thing. We didn’t check whether our early assumptions still made sense. AND Even the images, stories and maps we made steered our focus away from people’s capabilities and towards their victimhood.


Let’s look now at some alternatives to the narratives about intervention that are shaped by professional training and then shape our systems and procedures…. A lot of these narratives follow a formula that might be familiar wherever you practice engineering: people (especially powerless people) are not aware of risks, don’t learn, don’t know how to do a good job or are irrational.


Stories of Buildings

People don’t know about the risks….Haiti: Maybe this hillside is an excellent example of risk mitigation. It was a choice of location, from a limited set of choices, that minimised the risk of eviction, minimized the risk of high rents, minimized the risks and costs of finding and getting to work and it maximized the chance to live closer to acquaintances already in the city and it was a choice of building that was carefully adapted to protect from hurricanes, material scarcity, noise and burglary….

Solution: assume people are rational, risks are measurable and just take a broader view of risk…


Technology and imagination/imagined future

People don’t know how to build well: Mosques in Pakistan withstood flooding. When we sat with masons they explained and use of improved techniques, including:

Raised floor level;

Deep foundations, made from compacted earth (provide greater stability to walls);

Soil is sieved prior to use (removes large clumps and stones);

Walls are thicker, especially at base (enhances stability of walls);

Soil/mud is soaked overnight prior to use (disperses clay particles);

Wheat chaff or chopped jute bags are included in the mud render (reduces shrinkage cracks and increases water resistance);

Annual maintenance is undertaken.

And they had an incentive to take time over this for mosques in contrast to dwellings because the dwellings were occupied by landless tenant farmers and when they were moved on the home was raised to the ground and only allowed to carry with them their roof structure (which also became the “bank of mum and dad” where they put their savings - assets like girders poised on mud walls).

Solution: take a broader view of building knowledge accept that non-engineered materials can be made structurally and culturally appropriate and examine incentives of builders


People don’t learn: Padang Pariaman: this is a classic trope. It goes like this – structures aren’t braced so we are going to do some flyers and some training on the importance of bracing a structure. This picture is from Padang Pariaman in Sumatra, Indonesia. About 115,000 destroyed houses, 135,000 damaged houses. We got an innovative grant of £1m DfID funding for pilot of 3,400 households (3% of overall houses destroyed) and from the small seed of $330 people were given technical support to rebuild (not repair). This was targeted to rural areas and was successful – we said we would do this and we did this and it worked.

Settlement and Housing decisions:

•       Ownership (Mothers): Matrilineal System - women owning and inheriting land and housing;

•       Shelter Construction (Marriage): the new husband will move on to the land of his wife’s family; the wife’s room becomes the home of the new couple but when resources are available,

•       Site Selection (Family): new couples build an extension or preferably a new stand-alone house on the same family land, thus the priority in selecting a site for a new house is proximity to relatives and where land is owned (rather than vulnerability to physical hazards)

•       Shelter Modification (Births): people reported beginning with one room, incrementally extended on the birth of female children;

•       Shelter Finance (Jewels): on birth, girls are sent jewels, money and livestock and some people reported trading these gifts to mobilise additional cash to top up the DfID cash grant – one woman reported selling 8 gold bracelets to top up her grant; these assets also seem to be used in “normal times” to expand housing on marriage;

Majority of the people given the technical training were female householders (not because it was matrilineal but because they were at home – not out doing cash work - and had time to attend. But they had the least construction experience to get anything out of a technical training and had the least power – in spite of matrilineal system - to influence the decisions of masons – even the youngest, most inexperienced male masons who tended not have had any of the training.

[By the way, these examples are from rural areas and guess where NGOs did not go? Padang City: "a third of the households were located but with "a  disproportionally  low amount of shelter assistance". In April 2010, 54% of affected households interviewed continue to live in structurally unsafe houses, and that the majority of those in unsafe houses are economically vulnerable. Even in Padang Pariaman: Nice timber T-Shelter projects but people built like this and repaired their unconfined brick gables...

In Haiti this was different – if you owned land and could commission a building, you got to tell the builder what to do

Solution: think about things you can’t see and may not be able to know… power and voice and intersectionality


Mantra shelter is a process not a product - not just a product or a process it can be a service. And it can be an institution

I’ll mention to Kobe earthquake in a minute but what I want to think about here is notions of transience, permanence and displacement …. Originally the guidance on “transition” was not about a specified time period or a shelter kit – it was about the processes and trajectories of seeking accommodation after a disaster. But because pre-fabricated shelter kits are often called "transitional shelters" in our imagination it has become the kit or shelter object that is in transition – not people – and it reduces this to a technical problem of shelter design (e.g. can it be upgraded or moved later?). No further questions on people’s wishes are posed.

In Western Sahara choosing a place was about proximity to family (flood risk not on the agenda) and decisions about investment were based on an anticipated “return home” before consideration of winds and floods.

In Japan, outside major cities, land is expensive and the house is disposable… It depreceates in value more like a car than a property investment. Land is perceived to be “a permanent commodity” but houses have “an ephemeral quality” and are seen as transient objects that deteriorate, such that older houses become undesirably expensive to maintain and repair (Ronald, 2008), and some argue that this is possibly amplified by cultural and religious preferences for modernity and renewal (W. Johnson, 2007). In the disaster law, shelter after a disaster is seen by the government as a time-limited accommodation service – it is often a prefabricated shelter because the government prefers not to give hand-outs but to provide a service – they would rather spend 35k on a temporary house than give people 35k towards rebuilding. 

If your house is damaged, is it your chance to start over and build your dream thing with the compensation? Or make do and mend - patch up something that has sentimental value.

What I want us to think about is that a house is a kind of institution – shaped by us and shaping us – it is about agency – your choices in learning, adopting and sharing ideas; 2) the technologies of home architecture – inside and out – are about sharing of ideas, dreams and technologies by people  across time and space; 3) “change” is not just incremental, object-focussed, happens on a single path and ending with a stable,  permanent object (Fig 1) . It can be historical, it can be inexplicable (by us) or cyclic, evidence may not be within your grasp to explain how often or for what reason people make changes. Housing constitutes and catalyses wider change…

Solution: take a relative or contextual view of time not just space

Anatomies of Risk: What about our ways of thinking about risks?

Starting from scratch is less risky: So you might think this is a bit of a left-field example but there is a massive estate in Southwark called the Alyelsbury estate and while I was at UCL we did some work on the decisions to demolish social housing in London… We were approached by the London Tenants Federation and Just Space because they felt that “technical arguments” – i.e. that the O&M costs of these older buildings would be higher than those of the new equivalent - were being used to justify demolitions that were not socially, economically or environmentally justified. Apart from the evidence gaps – painted as “contradictory or ‘conflicting evidence’” rather [4], when actually is it was context-specific and patchy – and evidence impenetrability, the main problem was the pathetic standard of evidence: to demonstrate that the demolition was in the local public interest – economic, environmental and social interest  - which that is the basis for compulsory purchase – the planning application has to present scenarios. The technical scenarios are “calculated” as percentage changes or improvements.

But in an algorithm that optimizes for minimum impacts, the baseline against which you measure impact is critical. In the planning application – this is totally legal/normal by the way – the baseline comparisons are with a hypothetical new development that complies with current building regulations or with current “bare minimum” planning regulations. So there don’t offer any comparison to either the current situation at Aylesbury or a refurbished situation at Aylesbury.

This makes these figures largely irrelevant for making a rigorous appraisal of the environmental, social and economic impacts of the proposed redevelopment on the local area.

But it makes the figures look like they came from maths. And the thousands of pages of glossy planning documents – cost hundreds of thousands of pounds to commission from engineering firms.

Solution: when working on questions of viability, take a broader view of evidence (what is produced and by whom) and ask what evidence is missing, viability


Measuring risk accurately is important: so I interviewed one of the geotech engineers involved in the zoning of areas in Christchurch that were at risk of liquefaction. NZ is small – he’s bumped into the PM on the ground and they were totally committed to undertaking a thorough and expert assessment rapidly.

Some zones were obviously fine and some obviously not. But a few zones were hard to assess and this engineer was so concerned with doing a proper job that he wanted more time to do the work. He said that in retrospect he wished that they had classed everything as either red or green because the uncertain meantime in the amber zones resulted in huge delays, uncertainty, depression and suicides for people from those zones. He also confirmed what other interviewees has said that engineers working on reconstruction Erred on the side of caution: and this meant cowboys worked in difficult places, engineers only went to easy places (unlimited liability for residential work)

Solution: maybe it is best to be categorical (and a bit wrong) now rather than bespoke later? Or maybe the wider argument is that it is best to be realistic about all the other things that are happening…


Risk thresholds can be calculated: I’ve tried to have this conversation with so many engineers. In Japan after the Tsunami, the best data and most sophisticated models ever were used to simulate Tsunami flood zones and justify construction of a 500km sea defence – much to the chagrin of many coastal communities whose livelihoods depended on access to the sea. But if you ask why 1 in 100 return periods or 1 in 1000 return periods were used, the answer is often because of the modelling.

In the UK public health and health and safety use thresholds to decide when an investment is worth it. In health this is based on quality adjusted life years. So if a treatment can give you 1 extra year of high quality life for less than £23,000/QuALY, it will get investment. In health and safety if infrastructure – say a train – can be retrofitted to avoid a death for less than £1.3m you are obliged to do this. This is in a context of huge investment in evidence gathering and yet the CEO of NICE said to parliament:

Professor Rawlins admitted that the threshold was not based on "empirical research" as no such research existed anywhere in the world. He told us instead that the threshold was:

...really based on the collective judgment of the health economists we have approached across the country. There is no known piece of work which tells you what the threshold should be. Michael Rawlins, Chairman of NICE

A structural analogue…. Is the session I attended here on Eurocode 8 which was amazing experts largely volunteering their time to wrestle with the code

“Contrary to what managers, engineers, politicians and risk experts want to make us believe, it is the massive mobilization of the population, of dissident experts and of victims which have led ministerial departments, industrialists, safety committees and courts of justice to modify their attitudes”.

(D. Pestre, 2013, A Contre-Science. Politiques et savoirs des  sociétés contemporaines, p.151)

[Grenfell: comment on this horror]

Solution: acknowledge that the risk threshold is not a calculation but a deliberation


Standards should not be lower for poor people: The 1980s was the UN decade of water and sanitation but when the WHO looked at what it had achieved they had a massive internal debate about whether setting global water quality standards was the correct objective. Many argued that poor people should not be condemned to lower standards but others argued that setting standards condemned poor people to depend engineers - water quality to be regularly sampled, laboratory infrastructure to test samples and then – for micro-biological contamination – a 16 hour incubation time to check contamination…

Instead a whole new approach to systemic risk was developed which gave communities tools to check not the water quality but the worst hazards in the water system and then empower them to fix them.

A structural analogue to this might be the Italian code which has three different limit states are introduced: SLV (ultimate limit state (safety of people)), SLD (service limit state (comfort of people)), and SLA as limit state associated to damage to artistic contents (other values) (D’Ayala and Benzoni, 2012)

“upgrading... a specified safety level; improvement... a safety level higher than the current one, but not specified a priori; and local intervention... improving behaviour of specific parts…” Techniques for eliciting best technical consensus are next steps...

Solution: systemic risk in the aggregate, ethics and values



So. People had different explanations for what went down in Haiti: “we had hammers so every problem looked like a nail”. “we were used to designing complex buildings in simple contexts and could not get to grips with simple buildings in complex contexts”.

These questions remain live – the recent report by care and UCL on safer building acknowledges that these very questions “are hard to resolve and require substantial further interdisciplinary research…”

And I’ve deliberately set up these stories in a way that satisfies me as an engineer.

A diagnosis followed by a solution. Because I think some of the things that went wrong are to do with the way we think about the world, I’ve also deployed some brutal and unfair stereotypes about us as a professional group.

This is designed to stretch out the ways we think:

- So we take a broader view

- Try not to demonise, fetishise or fantasise about people in other places – whether this be estates in the UK or villages in Padang – in ways that delude us into thinking we can solve problems elsewhere

[Revisit Concepts of Shelter]

It can help us extend our concepts of shelter to include the ground, people and society not just as a collection of households, history and systems

[Revisit Algo]

It can help us try to extend our algorithms to include more things – although this is what happens if you try to draw out just a few of the things that were interacting in Kobe, Japan before and after the 1995 earthquake.

It gets messy quickly

But we will keep starting from an assumption that things are knowable and solvable and all that is needed is an extended framework, set of concepts or more factors in our formula

·      More measurements and evidence in our rational

·      More interactions and sub-systems in our systematic approach

·      More questions and iterations in our procedures

·      More humility in our professionalism

No more pictures or diagrams, diagnosis or solutions – I want to finish with something different: slippery ambiguous stories



Firstly, I hope I’ve shown that its complicated.

Just put some of the questions about repairing non-engineered dwellings against an engineer’s or a humanitarian’s code of conduct, and you will immediately find contradictions:  [“to undertake only those tasks for which we are competent, have regard to the public interest, not maliciously or recklessly injure…”; to do no harm, provide equitable help on the basis of need, build back better, minimise programme risks…

Secondly, I now think now that it was partly because of our pre-dispositions, our professional training and unusual power as engineers in that setting that the flow of money was drawn towards new stuff/delivering objects/engineered things and away from repair and fixing the things that existed and were crappy before and supporting the people who were already there.

We enjoyed importance and influence as technical specialists on the ground that would be unusual at home and generally we don't think of ourselves as simple "followers" of rules.

But our professional responsibility is NOT actually to decide what is "safe" – because that is socio-political, it is a deliberation not a calculation - our duty is to design and check for compliance (with safety rules). This was a near impossible conversation to have on the ground because the engineers I talked to framed this question of high safety standards as a professional/ethical duty to make something safer NOT as a professional dilemma about what their role in safety decisions should be...

So, I think it’s the contradictions that we have an ethical duty to confront.


Thirdly, there seem to be reasons why we don’t confromt these dilemmas. And that quote about hammers and nails always nags at me: should we not also think about the way a technical, professional education acculturates us and seems to encourage both an overblown sense of protagonism (thrusting desire to solve problems) and its twin tendency of [‘solutionism’] (preferring to abstract out of context only problems we are equipped to solve.

As engineers, we aren’t trained to embrace uncertainty. We also aren’t trained to think that expertise and power are distinct or that you could have one without the other. And at the start of an engineering career – especially in consultancy - we are not paying the piper or calling the tune so deciding what is built doesn’t come into it.

As people with an elite, technical and western education – we’ve had to learn to ignore some mistaken or naïve intuitions about the world – this is USEFUL. You might know that experiment where undergraduates are shown a film of a small ball and a large ball getting dropped from a tower and those without a physics background just cannot intuitively accept the film is real[1]. But studies also show that the scientifically literate tend to have polarized ways of seeing the world as opposed to people who may be less literate but more scientifically curious and remain flexible in their views[2]. We are taught to be analytical – we look at an object, find attributes, assign to a category, apply rules to predict behaviour. BUT studies suggest that in Japan – where learning is about social relations, harmony and an ability to coordinate action – students emphasise context, relationships between things and they can hang on to ideas that seem to be in conflict and cope with ambiguities[3].

In fact, it has been teaching that has given me a chance to think about how engineers are trained to think and how, for me, my education embedded tendencies to digest the world as systems and patterns but left me happily innocent about power. I wanted goodies and baddies, heroes and villains, comic book metamorphosis, measured by before and after photos.

I could see that I was naïve about the colonialist tendencies of the aid machine but I found a lot of the political critiques to be naïve about technology. Much as I came to hate the glorified garden sheds we dumped in Haiti, on the day the team witnessed the first shelter go up, we shared a euphoria.

Making something was like a kind of phoenix – a compulsion for the ordinary repair of a home that held such meaning in the rising after disaster.

And I didn’t want the written accounts to overlook the joy, the commitment, the mistakes, and the dignity in the human appetite for technological fixes and expertise.

Engineers come with this kind of unencumbered “solutionism”, analytical and careful skills, and often a willingness to improvise and innovate and – when we work together, and, blimey, I have been so so lucky with the lovely people I’ve worked with – we bring the humility of collective and anonymous authorship.

So I started to revalue the wonderful things engineers bring to the world.

AND as the recent press coverage about decolonizing Oxbridge curriculums has shown, just because you’ve had an elite education in a humanities subject, doesn’t mean you’re any more likely to be curious about plurality or the ways knowledge is produced….



Lastly, all this has given me to wondering whether our experience after disasters gives us both a glimpse into that future and a special insight into our profession. We work with juxtapositions worthy of SciFi apocalypse: epic destruction followed by the roll out of retinal scanning at a scale and sophistication beyond any technology we have at home.

We are on the cusp of some profound technological changes - some of which are likely to disrupt our [‘expert’ status as engineers].

We are entering the age of inference – where machines are not just doing but deciding. As Evgeny Morozov puts it in his book the folly of technological solutionism – data allows us to move from a world where we are obsessed about causality to a world where correlation suffice. Just because we have so much data. So you wouldn't need to go and build very sophisticated theories you wouldn't need to understand the root causes. just go and investigate how certain things are correlated….

Perhaps, with our practice as engineers, we can imagine the future of work as a fusion of craft and data. By craft, I mean being ‘on site’ making judgments about the sights, sounds and feel of material and paying attention to builders, users and place… Or – as a colleague once suggested – the skill to deal not just with complex structures in simple contexts but with simple structures in complex contexts. Marshalling analytical and diagnostic skills as well as material and human experience. Mastering the bespoke. Arriving at a shared understanding of risk before deciding unilaterally to mitigate. And making routine any bureaucratic or rule-based bit of design so that it can be done by an app or a cheaper human (elsewhere). But how many engineers are comfortable diagnosing unconventional materials and workmanship, non-standard details or severe damage? Will the people that pay expensive people like us to manage risk, keep paying for anything but the messiest, local diagnostic craft alongside the shiniest, remote machine learning?

By data, I mean more real-time, sensed data that might indicate (or not) the obsolescence of our stuff. Imagine that consumer protection and safeguards on public health & safety shift away from a system of codes, voluntary labelling and self-certification devised in ideal conditions and backed by governments. Imagine we shift towards real-time monitoring of vibration, strain or displacement – systems that can check performance in context. Imagine how this might challenge the ways engineers are regulated and self-regulate, especially given that we often come up with disaster ‘solutions’ that are essentially regulatory without considering the context in which such rules are created and policed. Would this make the dogmatic code-followers (and their lawyers and professional qualifications) obsolete and give new power to the lay-activist-geek-citizen? Would this allow risks to be seen in a broader context of trade-offs or over a longer period? What if we could ‘sense’ and ‘algorithm’ ourselves away from engineered regulation and towards public shame, sanction and deliberation?

What I’ve shown here are some apparently simple and discernible algorithms – open to scrutiny, written down, regulated by governments. What I’m trying to argue against is not engineering and technology but homogeneity.

But I want to leave you with this little story from another sector where the pioneers are trying to make – like Uber and air bnb - a mass-customised, bespoke product…in education:

“When the AltSchool technologists who participated in the December hackathon shared their discoveries at the end of the session, the team that had focussed on bookmarking video seemed particularly pleased with its innovations. The team had decided to try to find a “fun route” to help teachers request a video clip of a moment in class. “The idea is that the teacher could, in theory, just knock twice on their phone,” one team member said. He patted twice on his device, which was buried in the front pocket of his jeans, to demonstrate the ease and unobtrusiveness of the gesture.

Another member of the team tapped on his laptop, and a graph that resembled an echocardiogram, with troughs and spikes, appeared on a large video screen at the head of the table. A third team member, a young man with a starter beard, tapped twice on his phone, and the graph reappeared with a new spike—the result of his tapping.

There were cheers around the room as the developers explained how they had filtered the data so that the jostling motions of a teacher walking upstairs, say, would not show up as a bookmark. “It’s reasonably robust,” one said, with pride. Someone asked about a cluster of spikes on the graph. “That was, I don’t know—me digging around with the phone in my pocket,” came the answer.

From the back of the room, a woman spoke up: “Did you test it with a female?”

Many participants laughed. “I’m serious,” the questioner went on. “A lot of our teachers are females, and they carry phones in different places.”

The members of the bookmark team, all of whom were male, looked deflated. In coming up with their apparently elegant solution, they had not visualized a female teacher slapping her bottom to activate a phone tucked into her back pocket.

“That’s a really good point,” one of them acknowledged, his smile waning. “Yeah, it could use a lot of fine-tuning. This was just, like, hey, get ourselves to a demo.” They had failed fast and failed forward. That was what they were supposed to do. Tomorrow, they would iterate.”


These technologies will alter the status of engineers. Will I have a job in 5 years? Will I be more employable in 10 years as an engineer or as a STEM-oriented governess of an oligarch’s children? What if the world changes faster than our competencies? Precarious times. And fearful times, driving a politics hungry for the sanctuary of simple narratives, scapegoating and silver-bullets.

But my job is helping me to see how technology might answer the questions that plagued me after Haiti about our inability to invest in repair: I reckon instead of robots, riots and stalking – perhaps we – all of us - can make ourselves a world of craft, care and sensing. Enhanced by the engineering of fixing, getting dirty, addressing the bespoke and steering machine learning with other groups, people able to bring their experiences, qualitative, ethnographic insights to what causes things not just what appears, from a distance, to be correlated.

For now, what if we – professionals with some power over decisions – spent just 5% more time in any meeting – whether about cities, infrastructure, machines, algorithms – firstly thinking about who is speaking and who is silent and who is present and who is absent and trying to shift this we might rebalance decisions and change who the world is being built to please…




[1] Why we don’t believe in science? 7th June 2012, The New Yorker http://www.newyorker.com/tech/frontal-cortex/why-we-dont-believe-in-science

[2] The Problem with Facts, Tim Harford, March 2017 http://timharford.com/2017/03/the-problem-with-facts/

[3] Culture and Intelligence, Richard Nesbitt speaking at the LSE, April 2016. http://www.lse.ac.uk/website-archive/newsAndMedia/videoAndAudio/channels/publicLecturesAndEvents/player.aspx?id=3462