Lessons for Tech Architecture & Integration
I have the pleasure of having been part of a project, led by Fathm, where I get to coach news rooms on the opportunities and challenges of introducing new technologies, specifically tools connected to Artificial Intelligence, in their production workflow.
I now also have given a class, to two cohorts, one in Latin America, one in East Asia, on Tech Architecture & Integration.
Below, a reconstruction of the introduction to the actual class. The introduction dealt with personal anecdotes, leading to concise lessons, useful in the context of dealing with tech. I won’t bore you with the meat of the class.
I wanted to start with a brief introduction, going through a few personal anecdotal experiences that will work towards a few shared themes. Coincidentally, exactly themes that are relevant for the topic of Tech Architecture and Integration, in the context of AI.
I tend to present myself as a ‘creative technologist’ and ‘media artist’. But, before I arrived at that description of myself, back in the previous century, I received an M.Sc in Mathematics. I wrote my thesis in the department of… Artificial Intelligence. But, this was at the end of the second major AI winter, and way before AI research led to the current rise of LLMs.
I ended up working for Procter & Gamble, one of the largest consumer goods companies in the world, which was horribly boring, so I did what any sane techy did at the time, and moved to a small startup. But, we weren’t quite a tech startup. Our core product was predicting failure rates in maintenance processes using advanced statistics. So, we used statistics to predict the most likely outcome based on a large number of parameters… Not unlike what an LLM does today, though in a different context.
That was interesting. But then, somewhere in the year 2000, I found myself flipping through an old copy of Wired in a hostel in St. Petersburg, Russia, and came across an intriguing article on two early Internet millionaires who had decided to try and change the world for the better, using technology.
Long story short, this eventually lead to George W. Bush writing me a thank-you letter some 8 years later, and it fairly immediately saw me being shipped off to the country of Ghana, where I spent several months working with one of the nation’s larger news agencies.
This set me off on a trajectory that has seen me work with many news organisations in the Global South. And that brings me to the first of my themes for today:
Be flexible, find opportunity
In Ghana, I worked with JoyFM, a radio station, which also served the international Ghanaian diaspora in providing local news. Their offices in Accra had three floors. Just there, they had dozens and dozens of employees, journalists, radio producers, designers, editors, you name it.
And they had two IT staff. Whose primary responsibility was the hardware being issued by the organisation to its staff. And they were also responsible for the company’s website. And, they were not programmers.
In the four, or so, months that I was there, we made their website an award winner even if, in retrospect, the outcome was quaint; we made… a database driven website.
Well, yes, but we also put in place basic authoring, editing, and publishing, such that the journalists could manage their own content.
A decade or so later, I checked out their website, and could see, based on how the interaction functioned, that at least some of that initial version from ten years prior, was still part of their production environment. Primarily, because we avoided, as much as possible, introducing complexity.
This brings me to my second theme:
Keep it simple
Around the time I had a peek back at that Ghanaian website, I was asked to build a solution for the only independent news agency in East Africa. Somewhat similar to AP or Reuters, this agency (still) serves mostly Ugandan radio stations with news, much of which they generate themselves.
I spent a few months creating a fairly complex solution covering end-to-end pipelining, as well as tying the system into their financial processes.
But, I hear you think, surely, by the time I was to leave, ongoing support was going to be a problem. How were these Ugandans going to be able to pay for a support contract from some Digital Nomad hobnobbing around the world?
Well, the tech stack was fully self contained, while the product was pretty much feature complete. So, ongoing maintenance, since, has been minimal.
Sure, today, more appropriate solutions exist, but implementing these would require a serious effort, not just because of requirements around migration.
Which brings me to my third theme:
If it ain’t broke, don’t try to fix it
Somewhere in between, I also spent time in Sierra Leone. There, I worked with an organisation which had the huge responsibility of reintegrating war orphans into society. You might have seen the film Blood Diamond, with Leonardo DiCaprio, which is set during the truly horrid Sierra Leone Civil War, which, deploying child soldiers, resulted in large numbers of orphans.
These reintegration processes, literally, lasted years.
Now, all documentation, all paperwork, on years of facilitating these reintegration trajectories, were exactly that: paperwork, all stored in filing cabinets that no one could get their heads around.
Worse, this kind of administration was horribly susceptible to fraud.
I worked on creating a fully digital system which used an immutable database and cryptographic hashes to ensure data could not be tampered with. And this was a huge success; hundreds of cases of fraud were uncovered, success rates went up.
But it also showed that, perhaps, staff was underpaid.
Eventually, pushback against the system meant that it was dismantled after a few years. Certainly not because the system was not doing its job. Probably because it was doing its job too well.
And that brings me to my fourth theme:
Make sure you have buy-in from all interested parties when you make major changes to a system
Now, shifting gears, onto a more general perspective.
Some of you will have been in the news industry for long enough to have noticed the slow death of news media.
What’s the cause of this? Long story short: surveillance capitalism.
Specifically, particularly Facebook and Google first being the platforms through which news distribution was made frictionless, then expanding these platforms to replace the advertisement industry, made specifically these two platforms indispensable.
However, neither of them cares for ‘journalism’. They, and their shareholders, care about ever-increasing returns. And so, when both of them started funding journalism in the mid-2010s or so, they did not do this for a newfound love of news-production; They had realised that their monopolistic positions were killing off the Fourth Estate, and they were managing a politically dangerous relationship with journalism.
This has not ended. Specifically with the rise, and rise, of Trump, and Trump politics, journalists are starting to be referred to as ‘News creators’, equating them with ‘Influencers’, showing that the bottom line, for these Tech Behemoths, is views and clicks, by whatever means. And, they’ve got you over a fire.
How did we get here? Well…
Don’t put all your eggs in one basket
I’m currently coaching 2 newsrooms in Latin America, a third having left the program since it started. In previous rounds, I also coached newsrooms in Latin America, Europe, and in Asia.