Lessons from our US$500,000 supercomputing competition for Global South sustainability
A) Executive Summary
[Demonstration of See Through News’s focus on its Goal of Speeding Up Carbon Drawdown by Helping the Inactive Become Active and its zero-budget integrity/dynamism]
The See Through Carbon Competition:
- awarded US$500,000-worth of supercomputer-grade cloud computing
- to research projects that promote sustainability
- favoured Global South projects with no previous access to this technology
- Cloud computing management platform YellowDog donated the prize fund.
- Social media network See Through News (STN) conceived and ran the Competition.
- Social media network Global Sustainable Futures Network (GSFN) partnered.
- Dec 15 2022: YellowDog tells See Through News about donation of free $500K licence expiring April 10 2023
- Dec 15-31: See Through News partners with GSFN, assembles Expert Panel, designs Competition, creates online resources, develops marketing plan
- Jan 1 2023: Competition launches
- Feb 2 2023: 1st winner. Dr. Samuel Adekunle analyses emissions data from Africa’s 2 largest airports for carbon reducing action.
- Mar 15 2023: 2nd winner. Laura Degiovanni uses AI to detect sustainability Fake News.
- Mar 15 2023: 3rd winner. Johnson Jayeola creates unique dataset on firewood cookstoves in rural Nigeria.
- Mar 21 2023: 4th winner. Dr. Daniel Zepeda Rivas announced
- Apr 10 2023: Competition ends
- 2 projects completed compute by Apr 10 deadline: their research will be communicated to local people in STN-administered social media channels via STN-trained citizen volunteers, and will start measurably reducing carbon by the end of the year.
- 2 projects couldn’t prepare data in time: they remain approved pending new Competition.
- £0.00 (inc. tax)
Donations-in-kind (no cash) to make the See Through Carbon Competition permanent, bigger, and more comprehensive:
- Cloud computing capacity
- Cloud computing mentoring
- Cloud computing coding
B) Narrative Account
[Demonstration of the See Through News storytelling methodology]
An unexpected Christmas present
On Dec 15 2023 See Through News received an unsolicited donation.
A voucher, with an expiry date of April 10th, for a very particular product – supercomputer-grade cloud computing.
Half a million dollar’s worth of it.
For the previous few weeks, we’d been in contact with the donor, YellowDog, a cloud computing management platform, but this offer came out of the blue. How?
We first contacted them for an article we were writing about the Internet’s dirty secret – the carbon costs of the Internet in general and red-hot AI tech in particular.
YellowDog claimed to be using cutting-edge tech to pioneer a unique low-carbon method of cloud computing. See Through News was sceptically interrogating its claims.
We had good reason. Cloud computing is the biggest, and grubbiest, of the Internet’s dirty secrets. Cooling the computers in the data centres that power our Netflix ‘n’ Chills, Zoom calls, Amazon shopping and now ChatGPT noodlings, has a huge, and largely unacknowledged carbon footprint.
Calling bullshit on computing greenwash
As our article was intended to point out, the very term ‘cloud computing’ is greenwash.
The servers that drive the Internet and Big Data AI platforms require huge amounts of energy just to keep them cool – even when idle. Data centres are very much located on the ground, usually next to massive power stations that could service a small town, merrily pumping out greenhouse gases.
As data centres are usually hidden (do you know the location of your nearest one?), their emissions are less obvious than those coming from houses, planes, boilers, ships and cars. Most companies in this area of tech go to considerable lengths to keep it that way.
Like oil companies featuring wind farms in their ads, unchallenged, most cloud computing companies are pretty shameless about swanking about spurious ‘net zero’ or ‘green’ claims, and are largely permitted to get away with such corporate gaslighting.
Our article, AI, Climate Change and Monkeys Climbing Trees examined these contradictions, highlighting Internet greenwash and the hi-tech sector’s general reluctance to confront its rampant carbon output. Most people are surprised to hear the Internet generates around twice the greenhouse gas emissions of aviation, while attracting virtually none of its opprobrium.
Hence, our polite scepticism of YellowDog’s claims.
Not all cloud computing companies are the same
Despite being in the midst of a funding round, YellowDog CEO Tom Beese spent many hours over several weeks patiently explaining the advanced mathematics and distributed system behind YellowDog’s claims.
The Monkeys Climbing Trees article explains the tech in detail, but Tom convinced us of YellowDog’s claims to accurately and transparently measure the carbon cost of any compute.
YellowDog can’t compel any client to perform their computing task in the lowest-carbon way possible, but by providing a tariff for carbon intensity, as well as cost and speed, they offer clients the option to measurably minimise their carbon footprint, should they choose to do so.
Whether clients choose to take this evidence-based challenge to their claims for green credentials is between them and their consciences.
We’ve interviewed many computer scientists and hi-tech business people. Tom is, to use computer scientist jargon, ‘atypical’: a former British Army captain, lawyer, and historical fiction novelist, Tom is not your run-of-the-mill tech bro.
Over several long conversations, Tom not only explained YellowDog’s tech, but was a great help in road-testing and developing our metaphors for explaining the critical, but often misunderstood or underreported, AI technology that for better or worse, is shaping our future.
In the course of these conversations, YellowDog asked its own questions about See Through News. Tom was struck by our Goal of Speeding Up Carbon Drawdown by Helping the Inactive Become Active, and intrigued by the combination of our relentless focus on measurably reducing carbon over ‘raising awareness’ for its own sake, and the tricksterism of our ‘Transparent Trojan Horse’ storytelling Methodology.
Shortly after we’d published the Monkeys Climbing Trees article, Tom called to offer YellowDog’s half-million dollar donation.
Unwrapping YellowDog’s Present
Despite having just written a deep dive article on cloud computing, it took a while to understand exactly what the donation was.
Our ‘wow’ and ‘thank you’ were closely followed by the same four questions asked by others, when they heard about about the donation:
- What does ‘US$500,000 of supercomputer-grade cloud computing’ actually mean?
- What’s the catch?
- Why does it expire April 10th 2023?
- How is See Through News supposed to spend it by then?
1 – The Gift
Computing jargon still fresh in our mind, we asked YellowDog to quantify the gift with metrics we’d heard him use when explaining cloud computing. We asked questions like:
- How many GPU cycles does half a million bucks buy?
- How many petaflops to the yard?
- What combination of computing parameters?
- How many terabytes of data?
We had no idea of how to query their data quality metrics, like DF (Deployment Frequencies), MTTR (Mean Time To Recovery), or CFR (Change Failure Rate).
Just as well. In his characteristically measured, thoughtful way, Tom paused, before saying ‘Good question, but it’s so multifactorial, there’s no simple answer. Best just to explain it’s worth half a million dollars’.
At the end of this article, you’ll understand what he meant.
2 – The Catch
Still, everything we’d learned about the cloud computing business made us wary of this donation. This may seem churlish, until you understand the current business practice of most cloud computing giants.
They’re notorious for baiting new customers with ‘free’ or heavily discounted sweeteners, to capture their future business. Like Apple and Android phones, each platform has its own proprietary operating system, which locks clients into continuing to use their services.
Unlike banks and broadband suppliers, there are no government consumer protection regulations to oblige these giants to make switching suppliers easy. The first reaction of almost every computing expert we consulted about this gift was to assume it was one of these cynical commercial entrapment ‘special offers’.
At best, they assumed the cloud computing Big Boys were attempting to earn undue credit for dressing up idle capacity in philanthropic clothing.
At worst, they warned against getting tricked into committing to a supplier who’d start price-gouging once you had no option.
We’d explain that while this might be common practice, it wasn’t true of this donation. YellowDog has no servers itself. It has to source, and pay for, all computing capacity like anyone else.
This wasn’t the catch. The actual catch wasn’t a catch at all.
YellowDog licence came with one string attached – it had to be used to further the See Through News Goal of Speeding Up Carbon Drawdown by Helping the Inactive Become Active.
3 – The Expiry
Once we’d said wow and thank you, and had a rough understanding of the gift’s nature and purpose, we asked about the April 10 deadline. We’d never spent anything on supercomputing, but less than 5 months didn’t seem like much time to spend half a million dollars wisely.
But the deadline was immutable, because April was the end of YellowDog’s financial year. At the start of the year, they’d budgeted for a certain number of free licences, for projects that showcase their product. YellowDog had negotiated – and paid for – this compute time up front.
So the $500K value was real, and so was the April 10 2023 expiry date.
With Christmas imminent, this gave us a couple of weeks over the holidays to work out what to do with the gift, and just over 4 months to spend it.
This didn’t seem much time, especially as See Through News’ own budget is always zero.
4 – The Point
Once the first 3 questions had been answered, this fourth one was the most pressing – how could we use this donation?.
YellowDog was aware that though See Through News has its own AI projects, none is currently close to requiring the kind of muscle power they provide.
But they’d also become familiar enough with See Through News, and the integrity conferred by our zero-budget, volunteer-driven approach, to trust us to deploy its voucher wisely.
This was a humbling vindication of our work over the previous 18 months, but also presented a very real and immediate challenge – how to spend it?
Wrapping and Regifting YellowDog’s Present
The Christmas holidays were occupied with the very pleasant task of working out how to wrap and re-gift YellowDog’s $500K donation.
A few calls to our AI and computer science Brains Trustees clarified certain things:
- This was an incredibly generous donation
- $500,000 bought an awesome amount of number-crunching muscle
- Very few projects are likely to be ready for such muscle at short notice
- April 10th was a ridiculous deadline
- Especially ridiculous for a non-academic institution with no marketing budget, relevant network, or specialist profile.
The clock was now ticking. YellowDog had entrusted us with a significant commercial property, so the least we could do was to do our utmost to spend it effectively.
See Through News has a track record of coming up with original, creative, unique solutions at very short notice. Counter-intuitively, we’d found that not having a bank account can accelerate projects.
And so it proved in this case.
(Don’t) Show Me The Money
This revelation about money came as a surprise. Commerce teaches us to think of money as necessary fuel, or oil to grease the wheels. When you don’t have a bank account, money can magically start to look less like lubricant, and more like sludge.
Money has to be accounted for, audited, conferred about, deliberated on, agreed by consensus after careful deliberation. Seen in this light, it’s not strange that money’s absence means faster decisions and implementation.
The ‘only’ problem is how to get people to work for free.
That’s when a Goal of Speeding up Carbon Drawdown by Helping the Inactive Become Active come in handy. A much easier sell than toothpaste, tins of beans or documentaries.
Few disagree with this Goal, and many embrace it, seeing something new and different in our Methodology. Such volunteers contribute their time and expertise freely, figuratively and financially.
Not everyone can afford to work for free. For those who can’t afford to volunteer, you ‘just’ need to engineer win-win situations that create mutual benefit. This reciprocal benefit can come in many forms, such as:
- satisfying corporate ESG (Environmental, Social, Governance) requirements
- CPD (Continuous Professional Development)
- internships and work experience
- assuaging personal guilt
- wanting to do good in the world
Whatever the benefit, the fact that See Through News itself neither seeks money, nor has any, demonstrates integrity.
This is why YellowDog trusted us with a half-million dollar voucher. It’s why such a broad range of world experts in various fields (computer science, academia, carbon drawdown, communications, marketing, industry, corporate governance, accounting, to name a few) were happy to volunteer their advice on how to spend it.
We quickly concluded the fairest, and fastest way to share YellowDog’s gift would be to set up a Competition.
We called it the See Through Carbon Competition, partly because it sounded good, partly to stealth-promote a project currently under wraps, but which YellowDog knew about.
But who to target? Further conversations with computer science academics revealed a tricky Catch-22, which only became apparent once we tried to formulate rules.
Who to Target?
Our computer science Brains Trustees explained that outside of Silicon Valley Overlords and government agencies, the kind of computing power YellowDog was offering is only required by a certain categories of potential applicants.
- Established AI commercial companies
- AI start-ups
- Senior academic research groups
- PhD or Masters research projects
This left us with only one practical option:
1. we discounted because they shouldn’t need our help
2. should also have budgeted for supercomputing if they needed it, but might qualify
3. require high-level university approval even for accepting free gifts – an understandable precaution, but as these committees only sit 2 or 3 times a year, impractical for our deadline
That left only one choice as our key targets for See Through Carbon Competition applicants: postgraduate research projects.
Our UK and US university contacts assured us there were plenty of existing Doctoral projects that would fulfil the criteria of serving the See Through News Goal. Hard sciences like Geography, Physics and Biology addressed the Speeding Up Carbon Drawdown part, while Social Sciences and Behavioural Psychology address Helping the Inactive Become Active.
But this still left three major challenges:
a) Finding Applicants
b) Processing & judging Applications
c) Applicants being able to prepare their data by April 10
Or so we thought, until we started asking experts to describe the kind of PhD projects they had in mind in more detail.
Global North & South Needles
The more we listened, the clearer it became that the kind of projects our academic contacts had in mind, or could introduce to us to, didn’t really ‘need’ our help.
PhD researchers at top universities in rich countries with projects that required massive data crunching, would already have factored this kind of supercomputer-grade compute into their schedules and budgets.
They’d all be shuffling along the data processing queue, clutching their tickets for their time on the university supercomputers. Awarding them a share of YellowDog’s donation would,, at best, accelerate their research by a few weeks or months.
Whilst relatively low-hanging fruit, academic queue-jumping didn’t strike us as a useful way to spend this gift voucher.
So we asked about targeting PhD researchers in the Global South (i.e. the world’s poorest countries). Wouldn’t we, and YellowDog, get a much greater bang for our sustainability buck if we offered it to researchers with no access to their advanced technology?
Global South universities don’t have supercomputers, so researchers would never have the chance to exploit their level of computing power. Offering them YellowDog-level computing could make the See Through Carbon Competition transformative, rather than merely logistically convenient.
True in principle, said our advisers, but that’s the Catch-22 – without access to this technology, why would any Global South researcher create research projects that needed it?
The perils of gift horse distribution
Gift horse distribution turns out to be a tricker business than you might imagine.
One of the downsides of operating without money is that people tend not to value anything that comes free. They’re wary of scams. The more used you are to taking things for granted, the more likely you tend to be suspicious, and hence harder to convince.
More bluntly put, the richer and more privileged you are, the harder it is for See Through Carbon Competition to earn your attention.
This was another reason why focusing on Global South Applicants seemed like a good idea – the fewer resources you have, the greater the marginal benefit, and the more likely you are to grasp any free ones with both hands.
We’ll leave any reflections of how this relates to the causes and response to Global Heating to you, but it’s also crossed our minds…
But there was no time for philosophising. We’d learned a lot about cloud computing and academic research in rich and poor countries, but with Christmas already come and gone, we were stuck.
Our fallback was to offer YellowDog’s computing power to Global North universities. Our Brains Trust included eminent computer science academics from the likes of UC Davis, UCL, Imperial College, Southampton University and Edinburgh University. These institutions were warehouses piled high with sustainability needles. They just didn’t really need our gift.
With no money, and only 4 months to go, how could we sift through Global South haystacks for possibly non-existent needles?
And even if they did exist, would they be able to use this cutting-edge tech?
And even if they could, how could we find any by April 10?
This dilemma was resolved via the unlikely route of plankton-enumeration.
A perfect partner
A few months earlier, we’d written another deep dive article, about a citizen science project to count plankton in the mid-Atlantic, and the bizarre media maelstrom it produced.
After reading Dr Dryden & The Missing Plankton, one of the citizen scientists subscribed to the See Through News weekly newsletter.
A few weeks later, on her yacht in the mid-Atlantic, she read about the donation and the initial concept of the See Through Carbon Competition.
She subscribed to another newsletter from a different sustainability-related social media network, which she thought might find the Competition interesting. Bobbing on her yacht, she used some precious online access to forward our article about our plans to target Global South applicants, to this other network’s founder.
The recipient was Dr. Renuka Thakore, an engineering-trained carbon auditing expert. Renuka founded the Global Sustainable Futures Network (GSFN) around the same time as See Through News started.
GSFN is also zero-budget. It too was dedicated to measurably reducing carbon. It too had grown very rapidly, with 3000 global coordinators in 147 countries, giving it a reach of around a million multi-disciplinary researchers.
But GSFN took a different tack. It focused on early career researchers in the Global South.
Dr. Renuka Thakore was indeed interested. She got in touch, and our first problem was solved.
That’s when we discovered our next problem.
Setting rules in the dark
Renuka immediately offered the GSFN network to help find Applicants, and volunteered to provide a shortlisting service, should the volume of applicants require it.
So far as we knew, no one had ever tried offering supercomputing to Global South researchers before. Renuka’s guess was better than most people’s, and she thought there was a chance that a few Golden Needles might exist in the Global South haystacks.
See Through News and GSFN cobbled together a framework for the See Through News Competition, and ran our draft by YellowDog.
YellowDog had no formal veto, but would be involved in onboarding the projects,helping Competition winners prepare their data for cloud computing.
They’d factored in the demands this would place on their technical team, but targeting cloud-computing-novice applicants from the Global South meant this was likely to be much more time-consuming than any projects from Global North researchers.
Renuka warned that Global South applicants would require much more hand-holding than they might be used to, but YellowDog understood this was the cost of our ambition to regift their donation to people who’d never considered the possibility of using this tech.
Indeed, they were delighted at the notion of favouring the world’s poorest countries, appreciated the critical role GSFN offered in accessing the right candidates. They fully supported this unexpected new focus, saying this was exactly the kind of original thinking they’d hoped See Through News could contribute.
Excited though they were at the transformative potential of this approach, YellowDog’s technical team was already stretched. They pointed out that Global South projects were unlikely to take up much of the $500K, and though they were happy to split the award between multiple winners, there had to be a limit, especially if then all needed to be handled at the last moment. They didn’t want to risk offering a prize, but not being able to deliver it.
We were all in the dark about what the demand for this Competition might be, but agreed to mitigate this risk by setting a maximum number of winning projects of around 10.
Via our website, and GSFN, we announced the Competition, then the Competition Rules both in the form of a press release, and a ‘call paper’ format that might be more familiar to academics.
Ferraris for Peasants
Having GSFN as a partner gave us a fighting chance of finding suitable projects, if they existed. But Renuka also knew the issues confronting Global South researchers better than anyone we’d yet spoken to, probably better than just about anyone in the world.
Renuka bridged the gap between the world of high-performance computing, and the reality of the limitations under which Global South PhD projects are conducted. The more she told us, the wider that gap appeared.
Discussing the Competition Rules, another potential problem emerged. Even if we could find suitable applicants in time, might we be offering them a poisoned chalice?
To switch metaphors, were we offering a Ferrari to impoverished peasants accustomed to using ox-drawn ploughs?
Simply delivering the Ferrari, congratulating the winner as we tossed them the key, and leaving them to work out what to do next, was not being particularly helpful. Even if we could find candidates, was this learning curve too steep for them to stand a chance of generating any useful research by April 10?
Renuka’s insights alerted us to the need, within our zero budget resources, to provide a basic instruction manual and driving instruction, along with the Ferrari.
Renewing the collaboration that started the whole story, See Through News and YellowDog wrote and published a couple of online guides:
Exposing Rules to the Light
Before launch, we did everything we could think of to pre-empt or diminish known obstacles:
- We assembled an Expert Panel to assess the applications, explaining the procedures we intended to follow, and criteria we proposed to apply. Radical Transparency is a core See Through News value, so we were frank about the fact we were making the rules up on the hoof, apologising in advance for any inconsistencies, and asking for feedback to improve the process.
- We created an Application Form, balancing concision with clarity, making it clear we favoured Global South applicants with projects generating open source results that could be immediately implemented with no profit motive.
- We created a decent logo, thanks to volunteer graphic designer, branding consultant and web architects Watershed Creative
- We contacted a professional marketing agency, Goodwork, ‘the world’s first social enterprise marketing agency’ to help spread the word. Goodwork were inundated with work, but agreed to help promote the Competition as soon as their staff returned to the office in the New Year.
- We set the judging rules. Our system proposed each shortlisted application would be reviewed by at least one Technical Expert, who’d assess the project’s technical feasibility, one Subject Expert, who’d assess its academic rigour and robustness, and one Political Expert, who’d assess how well it served the See Through News Goal, and how quickly any results could start measurably reducing real-world carbon.
- We coordinated the formal Competition launch with Renuka. GSFN was ready to spread the word via its thousands of local coordinators in 1473 countries.
- We established a shortlisting plan, in case we were overwhelmed with applications. GSFN set up an emergency triage contingency plan, nominating volunteers to help shortlist applications if there were too many for See Through News to handle directly.
- We prepared our website, checking with our ISP that it wouldn’t crash if too many people accessed the Competition articles simultaneously.
The April 10th deadline loomed larger by the day. We’d done what we could. It was time to cut the ribbon for the See Through Carbon Competition.
On Jan 1, we announced the formal opening of The See Through Carbon Competition, crossed our fingers, and waited to see What Happened Next…
Waiting in the dark
….aaaand, at first, nothing much happened.
Renuka received some informal inquiries via her network. She directed them to our online resources and urged them to complete the Application Form as soon as possible.
A couple put in formal Applications, but they didn’t fulfil one or other critical elements of relating to our Goal, or being non-proprietary. We kept them in reserve, in case we had no choice later, but decided to hold out for an application that would tick more boxes.
Our first Winner would prove the task wasn’t quite as impossible as we’d been told. A good case study would be the best way to explain the Competition to potential new Applicants, and, in time, to potential new donors.
Goodwork worked on a press release. Our draft copy described the gift as coming from ‘an anonymous donor’. Goodwork suggested this might trigger some suspicion, which could be allayed by naming YellowDog.
Tom saw their point, but still preferred not to publicly identify YellowDog until and unless we had a winner to announce, lest YellowDog be accused of seeking unearned credit.
Goodwork distributed the press release, bearing the considerable cost of using specialist marketing email lists.
Hundreds of publications around the world published it, many in the Global South. A long shot, but another way to sift through all those haystacks for the elusive Golden Needles.
Until on Day 6 we received an Application Form from Dr. Samuel Adekunle, a Nigerian studying in South Africa. His project was called ‘Microscoping Carbon Footprint and Standardising Carbon Emission in Africa’.
Our first winner
Samuel’s Application Form informed us that for the past two years, he and his team had been gathering real-time emissions data from Africa’s two busiest airports and central business districts, Lagos and Johannesburg.
Samuel wanted to analyse this dataset to identify carbon-reducing actions that could be readily and immediately implemented on the ground, through social media ‘nudging’, in two of the Global South’s biggest and most dynamic economies.
Though conducted in collaboration with the University of Manchester, Samuel’s project did not have access to the kind of processing power YellowDog was offering.
This was our Assessment system’s first road test. We circulated Samuel’s Application to the Expert Panel. Three Experts responded with written follow-up questions, which we forwarded to Samuel.
- Our Technical Expert, YellowDog CEO Tom Beese, reckoned it looked suitable for YellowDog-level computation, though he doubted it would take a big chunk out of the $500K Prize Fund. Green Light.
- Our Political Expert, See Through News founder Robert Stern, thought it was a perfect fit for both elements of the See Through News Goal. Green Light.
- Our Subject Expert, Dr. Joe Jack Williams, had extensive experience of emissions sensor data collection. He had some penetrating questions on which he sought clarification. Amber Light.
Samuel’s written responses addressed these queries, but wanting to make sure of our potential first Winner, we invited Samuel to a video interview. He consented to recording it, in the interests of transparency and to help other potential Applicants.
A month after launching this seat-of-the-pants Competition, we announced our first winner, including an edited (and unedited, for transparency) recording of Samuel’s final video interview.
This was a huge relief to all the volunteers who’d put so much effort into what could have been a wild goose chase. It was proof Golden Needles not only existed in the Global South, but that we could find them.
Samuel and his team signed YellowDog’s simple contract. He confirmed he had the right to use all the data and software used to create it. YellowDog confirmed it would only hold the data for as long as it took to process it. We called it the ‘Sausage Factory Contract’. You hand over the ingredients, YellowDog transforms them into sausages, before cleaning the machines for the next client.
Samuel and his team started talking to the YellowDog technical team, who explained what was required to get their data into shape by the deadline.
We wouldn’t know for certain how much of the $500K prize fund would be used up, but YellowDog were very confident there’d still be plenty of the Prize Fund left over, so we kept looking.
The question now became – did we just get lucky, or were there more winners out there?
More Golden Needles in Global South Haystacks
Over the next few weeks, we found three more.
Laura is a UK-based Italian GSFN member. Her project Dynamic Map of Shared Environmental Truth, applies AI and Machine Learning to sift Real News from Fake News. This can be applied to any content, but she was particularly interested in sustainability and climate change-related misinformation.
Laura’s Application explained how her start-up TiiQu, ‘uses blockchain technology and artificial intelligence to disseminate, preserve from manipulation and make accessible the verified truth so that trust and collaboration can flourish’.
The Expert Panel concluded that Laura’s project can be applied to the first part of the See Through News Goal, Speeding Up Carbon Drawdown, and directly addresses the second part, by Helping the Inactive Become Active.
For reasons explained above, the Competition hadn’t targeted commercial start-ups, but neither had we ruled them out. Laura’s application was an excellent example of why. After written responses to our Expert Panel’s queries and a hastily-arranged video interview, the open source nature of Laura’s project’s won the day, and 3 green lights.
Laura’s team were next to contact the YellowDog team for onboarding.
Johnson, GSFN country coordinator for Nigeria, is an industrial chemist by training, but as we were to discover, he not only wears an impressive array of different hats, but is a millinery marvel, prepared to create entirely new hats, at the drop of one, from scratch.
We’d assumed it would be a stretch for Applicants to adapt existing projects in time for our deadline, but Johnson’s Application defied our expectations by creating an entirely new project in response to the Competition.
From Firewood to Zero-Carbon Pellets: cookstoves to reduce carbon and improve health for rural Nigerians, clearly ticked the Subject and Political boxes. According to Project Drawdown’s Table of Solutions, Clean Cooking ranks as No. 6 on the list of the most effective things we can do to measurably reduce carbon,
Having tested The Learn Game, a gamified Table of Solutions, on dozens of people, including at COP26, we knew even the world’s most active climate activists were unaware of the carbon cost of firewood cookstoves, not to mention their biodiversity and public health impacts.
The Expert Panel gave Johnson three green lights, only three weeks before the Competition deadline.
Johnson assured us he could design the project, create an app to run it, rustle up a volunteer army of field surveyors to collect data from all over Nigeria, and prepare the data to meet YellowDog’s demanding specs before April 10th.
Daniel Zepeda Rivas
Daniel, a London-based Mexican architectural engineering researcher, was the only winner not to come via the GSFN network. Expert Panel member Dr. Joe Jack Williams has been collaborating with him on his project, Zero Carbon Buildings for Changing Climates.
Having encouraged Daniel to apply, Joe Jack recused himself from the approval process, but as soon as the rest of the Panel read Daniel’s application, we realised why he’d recommended it.
Daniel had assembled a huge amount of data from 500 global locations, to measure the carbon intensity of existing local construction methods, and identify how zero-carbon replacements could adapt new buildings for our changing environment, with more extreme temperatures, flood risk etc.
Given Building accounts for 39% of historic greenhouse gas emissions, this triggered three green lights from all Experts. YellowDog’s technical team were particularly impressed and excited at the volume of Daniel’s data, and the complexity of his computation.
But getting the green light only 2 weeks before the April 10th deadline, could Daniel and his UCL colleagues whip this huge amount of data into shape in time?
The diversity and creativity displayed by the four winners was a huge vindication of what we hoped would be not a one-off, but serve as a Pilot for the See Through News Competition.
We hoped finding four such remarkable case studies, with no budget and a ridiculous deadline, would convince other donors to contribute to a permanent, more comprehensive Competition.
In this sense, the Competition had completed Proof of Concept. Our four Winners all said very nice things about the Competition:
- ‘monumentally inclusive’
- ‘impressively transparent and efficient’
- the process was ‘distinctly characterised by openness and honesty’
- our Expert Panel had ‘helped to crystallise the narrative through insightful questioning’
The Final Furlong
The race was now on to see if our four Winners could prepare their data in time for the rapidly-approaching April 10th deadline.
All four winners were going right up to the wire. Given they were cloud computing novices, this was expected, but YellowDog’s technical team, though expert, was not large, and had commercial clients too.
Nor could we be certain that the U$500,000 free licence would cover all four winners. Until Daniel’s project was approved with 2 weeks to go, YellowDog was confident there’d be plenty to spare, but now they’d seen the scale of data he’s accumulated, they weren’t so sure. We hadn’t come up with a plan for this contingency.
Our computing experts had warned us that the kind of computing firepower offered by YellowDog was a completely different proposition from what our Applicants were used to, by several orders of magnitude.
Comparing cloud computing to Global South university servers wasn’t a matter of attaching an electric motor to your bicycle. It was more like upgrading from a bicycle to, well, a Ferrari.
Without a driving instructor, manual, spare parts maintenance, fuel, parking and chauffeur, the risk remained that our Global South winners might be better off using their bicycles, in the short term at least.
YellowDog’s technical team could help them understand what was required to make use of their platform, but not to do the work itself. Depending on the state of the raw data, this might require considerable bespoke data tagging and additional coding.
For the moment, all the Competition Prize Fund could offer was YellowDog’s Dec 15th donation.
Essentially, here’s your Ferrari, good luck.
Even gift horses need to be fed and stabled. Would our 4 winners make it under the April 10 wire?
What Happened Next?
Two did. Two didn’t.
Defeated by Deadline
Samuel and Laura had to admit defeat for different reasons: she was subject to the competing demands of being a start-up in the middle of a Venture Capital funding meltdown, he to the demands of managing university procedures.
But fundamentally they were vanquished by the same implacable foe, Time.
With limited resources and the Competition’s out-of-the-blue deadline, Samuel and Laura realised the amount of additional work required to meet the April 10th shut-off was more than their limited resources could afford. Their cost/benefit analyses concluded they had to prioritise pre-existing demands.
This was entirely understandable, indeed what we’d anticipated.
We told Samuel and Laura their projects remain green lit. We’re now trying to make this Pilot project permanent, and bigger. If we succeed, they’re back in the queue, only this time, we hope, without the unnecessary stress created by the April 10 expiry date..
Driven by Deadline
But deadlines sometimes work in your favour.
Daniel called in a huge variety of favours from all corners of University College London. Various expert colleagues had agreed to check his hundreds of different data sources, to ensure they conformed to a consistent tagging standard.
The April 10th deadline magically levitated those promises from the bottom of UCL in-trays to the top.
Serbian coding wizard Dr Ivan Korolijia is in great demand at UCL. Once Daniel explained the peculiar terms of the Competition and its potentially transformative impact on his research, Ivan dedicated long hours to the project, providing great help and support .
YellowDog called what they came up with ‘a perfect fit’ to showcase their computing muscle power.
High performance computing experts, it turns out, dream of applying 708,588 iteration files of detailed building characteristics to weather conditions from 500 locations, resulting in 354 million discrete simulations in the cloud on AMD EYPC E4 Flex VMs (virtual machines) with 64 OCPUs (128vCPUs) and 64GB RAM, consuming 45,000 instance hours of compute concurrently across five cloud regions in North America and Europe over 11 calendar days, with less than 0.00003% failure rate (automatically and successfully retried), creating 37.5TB of data.
You may remember Tom’s suggestion to stick to ‘half a millions dollars worth’ when attempting to explain what was on offer. Now you know why.
The carbon cost is still being calculated, but YellowDog expects it to be the industry standard, as the deadline was too tight to deploy its carbon-saving tricks, like selectively distributing the computing tasks to renewable-powered data centres.
As for how much of the $500K prize fund Daniel’s project used up, that’s still being calculated precisely, but YellowDog reckons it was probably in the region of $100K. When you consider the complexity and scale of Daniel’s project, this shows just how generous the donation was.
Daniel describes the value of the donation as ‘immeasurable’, saying it ‘changed the course of the research project’. He says the See Through Carbon Competition enabled him to ’compose a more complete and detailed dataset’, the output of which, once analysed, will be ‘useful for many people around the world’.
Johnson’s compute, by contrast, barely caused the YellowDog dials to twitch. Then again, Daniel had accumulated his data set over years, while Johnson’s team of 45 young Nigerian volunteers collected it the week before.
Johnson’s project may have been less of a technical challenge, but we hope it will give potential future donors to the permanent See Through Carbon Competition pause for thought.
There’s a huge reservoir of dammed-up Global South talent. The See Through Carbon Competition can be the conduit and turbine that turns this potential energy into a force shaping a sustainable future.
For donors who’d like to be part of the solution, it’s an opportunity to support the vision, energy and capacity to make things happen, of people like Johnson. Open the door a crack, and they’ll grab the chance to discover new realms.
The creativity and drive Johnson demonstrated in cooking up his project from scratch to supercompute in a matter of 3 weeks is an astonishing vindication of what making advanced technology available to the Global South can achieve.
The dataset, facilitated by this Competition, is unique. Johnson will make it freely available, including to the Nigerian government, working with decision-makers at national and state level, to help direct and measure their carbon-reducing policy.
Like the See Through Carbon Competition, Johnson sees the achievements of the past few weeks as a Pilot project, kickstarting something of much broader significance and impact.
Now he’s experienced what the application of Global North tech to Global South sustainability can achieve, he wants to expand his cookstove project within Nigeria, and beyond.
Once the model is established and fine-tuned in his native Nigeria, Johnson wants to replicate his open-source project throughout Africa, and anywhere else in the world where people still use firewood to cook.
What Happens Now?
Not long after Tom informed Robert of YellowDog’s extraordinary donation on December 15th 2022, they started speculating about where it could lead.
The See Through Carbon Competition was always intended to be a Pilot. As the project developed, our goal became more specific – it was a proof of concept, testing whether projects that could lead to measurably reducing carbon in the Global South actually existed.
That question has been answered with a resounding Yes. Now it’s over to any potential donors to follow YellowDog’s lead. We welcome others to make similar no-strings attached donations, placing their trust in See Through New’s transparency, and their faith in the talent of people like Samuel, Laura, Johnson and Daniel.
The Pilot Competition has confirmed what’s required. No cash donations – See Through News still has no bank account, so we couldn’t accept money if you tried to stuff it in our pockets.
What we do need is more computing processing power committed over a longer term, and experts to help mentor, project manage and code future winners.
We need more Ferraris, a pool of driving instructors, and volunteer mechanics, to help inspirational people like the Pilot Competition’s winner steer us towards a sustainable future.
Johnson and Daniel have volunteered to join any future Expert Panel. Their hands-on experience will be invaluable in helping others go through the journey they’ve just trailblazed.
They’ve also volunteered to join a new group of See Through News Country Coordinators, part of an in-house, zero-budget, pipeline to:
- Create carbon-reducing solutions (via the See Through Carbon Competition)
- Train local citizen journalists to communicate them (via the See Through News Global Reporter Intensive Training (GRIT) scheme
- Acquire unique access to local Unwilling Inactivists (via the See Through News Facebook Group Network)
There’s no shortage of motivated, passionate volunteers who want to help Speeding Up Carbon Drawdown by Helping the Inactive Become Active.
Whether you’re a global corporation, a school, a small business, a grizzled veteran or a novice, you’re most welcome to join us.
If you’re interested in volunteering for any See Through News projects, please check what’s on offer at www.seethroughnews.org (everything you see has been done by volunteers over the past 2 years, on a budget of $0.00, including tax). If you’d like to join the team, complete our volunteer form, and we’ll get in touch with you to find the best fit for your experience and goals.