Hi everyone.

The world continues to deliver headline upon headline about generative AI, and I feel like Mickey Mouse trying to mop up the spill in Fantasia as more water gets poured in by the enchanted brooms. 

It’s not that I don’t enjoy seeing our fave in the news, but we’re now moving to this part of the news cycle that’s trying to spin things so far into the future that it’s just a projection of our fears rather than a projection of what the technology can or would do. 

“Will generative AI put all journalists out of business” is like a parent writing 5,000 words about whether their kid has the right assets and capabilities to find the cure for cancer when she grows up. But also, the baby is two weeks old. 

So, I promise we will return to this topic. In fact, I am brewing a report that deals with NLP. So if you’re working on NLP at your org and want to be featured in this report, consider this your bat signal. 

This week, for a break, we’re going to talk about data careers.

All my best as always, Ariane

[BUTTON: CLICK TO SUBSCRIBE]

PS: Your INMA membership gives you access to all our community Webinars, and we have one coming up on February 22 with David Caswell, executive product manager at the BCC, to talk about innovation and infrastructure for audiences, automation, and AI. Sign up!

An early career as a data person: transforming media companies and working on diverse problems

In these parts, we’ve looked at how the mission of our companies — informing society — was often a big driver in candidates applying for roles in our organisations. We’ve also looked at the changing expectations of data folks in terms of work-life balance and lifestyle (read: they want to be able to work remotely, and you’re not getting them back in your office building just by making it law).

And this is one of the reliably delightful parts of getting to chat (as a job!) with smart data folks across media organisations — being mission-driven is a fuel that makes people passionate about the job they do.

I’d like to make the generalisation that all data folks at media companies are like this, but I am reminded of that time where I observed to a Canadian friend who had never met an unfriendly Canadian person. So it didn’t seem there were unfriendly Canadians: “We don’t let them out of the country and they are not allowed to speak to foreigners,” said my friend. Which is another way of illustrating that I should heed the possibility of a selection bias in my sample.

But speaking to an early career data scientist from a large Benelux publisher last week, she pointed out a key driver in her own decision to join a media company straight from university: The desire to change the company itself and change the data culture specifically. Another early career data professional pointed to the diversity of problems to work on as a motivator. 

“I want to help journalists get the information they need,” said Roza Dorresteijn, a data scientist at DPG Media. “They shouldn’t feel judged by data — our reading of data should always be in service of the mission to inform,” not in service of building a business strictly for dollars and cents. “What makes me so passionate about establishing a healthy data culture within the publishing company contributes to the outward mission of independent and high-quality journalism.”

This echoed a conversation I had several months earlier with a data manager who works on the commercial data side of a large UK conglomerate — someone who didn’t work with journalists. But they saw the mission to make the company more data savvy as an important motivator in their work: a stronger data culture as a way to underpin transformation. 

When you read job descriptions for data roles at media companies, you reliably find mentions of the company’s outward mission — its societal purpose — as a reason to join. HR teams do understand that folks work on the technical side of media organisations because they want to contribute to the mission to inform the company. But they don’t often tout the motivator of transforming media companies themselves. 

Meanwhile, another motivator of early stage folks is also the diversity of data problems at media companies: subscription, advertising, user engagement, product data — and crucially there is usually lots of data. In fact, it’s the relative abundance of the data relative to our use of it that itself provides a lot of ground for the early career folks. 

Our UK commercial data manager was able to push through new concepts and programmes because, simply, they saw an open opportunity and no one was doing anything about it. Their manager supported their plan, and off they went.

Meanwhile, our Benelux data scientist was hired to build a new practice that simply didn’t exist before. So while all the ground is fresh and new by definition — and this could feel a bit daunting — this is exactly what an early career person wants.

A senior data scientist’s view of media company careers: look to Big Tech best practices

Now, an early career data professional at a media company will eventually turn into a more experienced data professional — someone who’s been around the block a bit. And we want to keep these folks around because there is a lot of industry-specific context to our work.

So while bringing in fresh blood from outside our media industry is always great for the new vantage point we bring to our shores, we should want to cultivate our own people, too. 

As this professional matures, their needs and perspective on their career in media data also changes. Speaking with a senior data scientist at a preeminent German publisher just recently, they pointed out a number of areas with challenges for media companies looking to hold on to their data team members:

• Career banding being oriented at management for progression rather than allowing a longer individual contributor track.

• Communicating about the work the company is doing in data — both as a way to support team members thought leadership but also a way to connect with other folks working on similar problems

The career banding issue is actually one that Big Tech essentially solved several years ago — recognising exactly the challenge raised by our senior data scientist: “I want to stay on the technical side of things because it’s really what I enjoy and I think that’s where I’m good at so I’ve seen a lot of people moving away from actually writing code to managing projects.”

Meanwhile, “for many media companies, you don’t have something like technical career development. Most of the people go from like, junior developer, senior developer to just, well, management.”

In Big Tech, you will often find two parallel career tracks: One of the tracks heads into management, and another track is for individual contributors (IC) —  with principal engineer type of roles. Google has two roles in its IC track with lovely titles: “Distinguished engineer” and “Google Fellow,” which sound very elevated but do highlight the scholarly, rather than management, nature of the titles.

The whole design of the IC career track was to allow folks like our German data scientist to continue progressing in their career — taking on more critical engineering roles where their expertise would still be applied in the context of being a builder rather than a manager of people.

As data gets into a more and more specialised game, you do want to allow folks to continue to specialise and acquire extreme expertise. So it becomes more urgent to structure data careers to allow this in the first place, where progressing (making more money and having access to the most ambitious projects) isn’t requiring your collaborator be forced to move to management. 

Which doesn’t mean that mentoring younger team members doesn’t appeal. In fact, the drive to communicate about both expertise and work is another big motivator:  

“We’re getting younger people, and I’m really interested in helping them grow and also grow as a team,” but this mentorship doesn’t necessarily mean management. Skill transmission is as much training, brown-bag chats, and good team knowledge transfer as it is about having a boss who is nominally encouraging you to stretch yourself in your current role on your way to your next role.

Looking outward to the wider community of data folks — in media and outside of it — there is a question of how to build bridges. Also on the mind of our senior data scientist: “Before, there was Twitter and people used to be out there. Now it’s unclear where people are congregating,” making the need to outreach more formally — with blog, conference participation a more acute goal. (Shameless sidebar: It’s me, hi. If you want to speak at INMA events, please drop me a note.) Something where our data scientist would hope to see more encouragement from their company.

As it happens, this outreach can be a virtuous cycle — not just for data team members creating stronger networks and peers to draw knowledge from. It can also be excellent for recruitment. A few months ago, I casually mentioned a Medium post from a New York Times data scientist to someone in the data organisation at The New York Times. I said that I probably hadn’t managed to grasp all the finer points but found it super intriguing. Yes, this person said, and they had very much figured it would be intriguing for the kind of candidates they would hope to engage with — the post was out there in large parts to create excitement in the data community and hopefully get some applications for some of their open roles.

Further afield on the wide, wide Web

Filed under “I see what you did here,” an interesting, out-of-left-field way to explain a look-alike audience (or a data proxy): The New York Times looked for ways to estimate the death toll from the recent COVID spike in China, which may have been under-reported by the Chinese government. So The Times looked at obituaries published for the researchers and scientists of several key Chinese research institutions and academies, whose affiliates are often of more advanced age. This a fascinating, and ingenious look, at reimaging sampling for a different storytelling use than what was intended.

Filed under “Obligatory AI round up of links”:

• The Washington Post tells the story of Big Tech’s contribution and reaction to the arrival of generative AI tools. This includes the perception that as folks get excited by the tools that are out there, large tech companies are forced to both hasten their own releases but also consider the impact of these releases perhaps in a different way than the smaller companies that have already moved in the field. (Article is a gift link) 
• The New York Times looks at history of the release of OpenAI’s ChatGPT (Article is a gift link).
• Just this week, Google announced its own generative AI chatbot, Bard (The Verge), but also a new library to generate music, called MusicLM. 
• While Meta announced a tool to turn a regular video into a 3D version by using text (Twitter thread from Ben Tossel). That said, great news about MusicLM, but she needs to know my own father built a programme to generate a Bach cantata based on all existing Bach cantatas, and this was in the early 1990s.

Filed under “We hadn’t heard from you for a minute,” some updates about the Cookie Apocalypse: FLEDGE, is edging out Topics in Privacy Sandbox tests from publishers and other industry groups (Adweek).

Community Webinar coming up February 22

We’re gathering with David Caswell, executive product manager at the BBC and former lead for its News Labs to discuss the building blocks of data-driven innovation, from infrastructure to organisation. I’m super excited to hear from David, who is a font of knowledge and experience. Free to all INMA members, so do sign up. 

About this newsletter

Today’s newsletter is written by Ariane Bernard, a Paris- and New York-based consultant who focuses on publishing utilities and data products, and is the CEO of a young incubated company, Helio.cloud. 

This newsletter is part of the INMA Smart Data Initiative. You can e-mail me at Ariane.Bernard@inma.org with thoughts, suggestions, and questions. Also, sign up to our Slack channel.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .