They are not talking about writing "glue code" (misleading title perhaps). They are talking about engineering management and senior responsibilities:
> You'd like to have time to code, but nobody else is onboarding the junior engineers, updating the roadmap, talking to the users, noticing the things that got dropped, asking questions on design documents, and making sure that everyone's going roughly in the same direction... If this describes you, congratulations: you're the glue.
Edit: Looks like the title was changed to "Being Glue" which is probably a better title.
It's only really in the last year, maybe two, that I've become comfortable with this being my role. I'm a jack of all trades, so I fill in the gaps within my team.
About as much of my job is about ensuring everyone else on the team is able to be their most productive, as it is applying my skills and experience to a project of my own. Sometimes that means working with engineers in my team directly on stuff, or acting as a sounding board. Other times that's getting involved in certain meetings to be able to present a more detailed technical response, and potentially head things off at the pass. As I've learned to take a step back from things, I can see just how much my influence has effect within the group.
I do still love getting my hands dirty, I love solving problems. I've no intention of ever stepping away from that kind of work. I don't really have to, either, even with the seniority.
I like the point she makes at the end... if you don't know how to do the glue work, communicating effectively, onboarding others, etc... you're not well rounded... and shouldn't be promoted until you learn those skills as well.
I think the term "glue work" somewhat hides the importance of building strong professional relationships and being able to develop as a human within your peer group.
That ability, though, hinges entirely on the affordances provided by an equitable social context. That is: mutual respect, acceptance, understanding, open mindedness and so on.
In that regard, the earning the title "senior" isn't just a reflection of your own maturity, who is awarded that title is just as much a reflection of workplace culture, it's values and the available opportunities.
This role is often referred to as "Product Owner".
If you have a senior engineer doing the work described in this article, then they have stepped into a Product Owner role. The only real problem with that is when it is done accidentally. If they want to continue to code and advance their technical career, this isn't the right move. If they want to move out of coding and into leadership, this is a good move.
Either way, deliberately acknowledging that they are now a PO will help focus their efforts, as there is a wealth of information online about doing that role well.
I agree. Having flexible but well understood roles is key for good team management. This seems to be lacking in many orgs.
Product Manager: responsible for technical and commercial product-market fit. Getting Voice of Customer. Understanding market size. Setting pricing.
Electrical/hardware/software Engineer: technical producer of the code/hardware/etc.... unit that is used in the product.
Systems Engineer: responsible for ensuring requirements of the project and product are met, tested, and giving confidence to the business that they have tested product, that will meet the user needs, will work in the environment where the product is used, and is ready for production.
Product Owner: Ultimate decider for product features in real-time. Ensures during development that the right product is being built, everyone knows what should be known.
Core team lead/ Project Manager: the person enabling the project to proceed. Manages business, timeline, requesting headcount, enabling critical path to move forward, etc...
Engineering Manager: responsible for hiring and training engineers, ensuring engineers have appropriate workloads, ensuring engineers have desired career development.
Depending on the size of the team and project, people may wear more than one hat. In many software shops they usually just split out Product Manager, People Manager, and Software Engineer. Thus the "glue" pieces are those roles missing above.
This is accurate, not for me but for another person on my team. He struggled for years to get promoted because of a perceived lack of focus from upper management. In reality he was handling a big part of the TL's job and several other people's job. The happy part of the story is that he got promoted a few months ago.
I am someone who enjoys/excels at being a "glue", I'm not quite an engineering manager yet but I don't write tons of code anymore either. I have been struggling to find a job. Every job I see is either entry level individual contributor or senior manager. Where do I find companies that value and seek glue/leads?
I don't know about job ads you've seen, but senior manager is often a dream/wish of companies who'll hire the first person that won't make them facepalm during the interviews after 6 months of not finding the right senior.
> Where do I find companies that value and seek glue/leads?
I think most companies value such work; it's just important to find the language they use. For example, in my BU, we have agile scrum masters. They are doing essentially the same work as technical project managers in my previous BU from the same company. I'd just advise reading the ad language and deciding whether that's "glue you like" or not...
Get in the door as IC and then gradually and informally be social, ask to be in meetings, and do "free" glue work. If you're good at it and it's valued, over a few months you'll naturally be spending more time on that and slowly be assigned fewer and fewer IC tasks, after which you'll have a conversation with your manager about which direction you'd like to take, and then you're defacto glue even if that's not your job title. (I'm sure you could get a job title change if you wanted, but SWE may be a slightly higher pay scale than TPM. ;)
N.B. this advice is really for "normal" times; if you're remote it may be harder or easier to do the social part necessary, I have no idea.
On my team, this glue work is the responsibility of the Product Owner.
Essentially, it is the PO's responsibility to ensure that everyone on the squad knows what to do and how they should be doing it. If not, the PO calls a meeting and brings the right people together until a consensus is found.
Having a full time PO on each squad lets the PO take up any work that isn't coding, but is blocking squad progress.
This post might as well be describing me to the point that I feel both happy and sad.
It feels so “wrong” to me to not share knowledge that I know can help someone else on my team immediately to the point that it becomes difficult to do what I am supposed to do.
I don’t know what I should do. Is there a book or a video that I can learn from? Maybe a mentor?
My dog's name is Dotcom and I bought him the domain dotcom.dog.
So, naturally I immediately concluded your link was going to be about a dog as well. Before clicking I even convinced myself that 'noidea' was most definitely your dog's cool and funky name.
Great talk - I have a feeling she'd get on well with rachelbythebay, whose blog has a lot of posts on the same theme.
The sad fact in some places is that as a woman, your choices for skirt length are "too prude" or "too slutty", and your choices for the amount of glue work you do are "not technical enough" or "not a team player".
Some of these places then act really really surprised when their women leave - sometimes leave the field entirely.
My humble contribution of the day for managers, if you've tried everything else and your equality metrics are still not going up:
1. Who is the glue keeping your team together?
2. Promote them. (Not move them into management, promote them to your equivalent of senior engineer.)
> Promote them. (Not move them into management, promote them to your equivalent of senior engineer.)
I think this is controversial. If you have a software engineer being promoted to senior software engineer (1 -> 2) without at least intermediate knowledge of the codebase, that'll make other seniors suspicious/wary.
Personally speaking, if you hire a software engineer that does mainly/only glue work, you mis-hired. What you meant to hire was a technical project manager. Move that person to a TPM if they want to (and are great at what they do) instead of having a senior software engineer who hasn't really shipped anything noteworthy.
> If you have a software engineer being promoted to senior software engineer (1 -> 2) without at least intermediate knowledge of the codebase, that'll make other seniors suspicious/wary.
She clearly did. Without that she would not be able to onboard or contribute to the design discussions and would not be able to help customers out. You can't assume that customers implies non-technical. In mid to large orgs most of your customers are other teams with very technical needs.
> if you hire a software engineer that does mainly/only glue work, you mis-hired
No, it depends on what your team needed. You can't have a 5-person team where everyone is great at crunching code unless your only success metric is LOC produced. You need diverse set of skills to deliver and maintain the product. You mis-hired only if your glue competencies are already covered.
In general I see this as a massive problem that many engineers believe that technical excellence is all that's needed to build a product. It isn't. Design reviews, code reviews, and onboarding for technical staff are very much technical.
And simple technical tasks that are tedious and boring are still technical tasks. And as a senior I would never assign them to a junior developer unless there's a lesson to be learned by doing them (once). If your glue person is picking them and you judge glue person's output by the complexity of those tasks (not technical enough!) then your metrics are wrong.
> She clearly did. Without that she would not be able to onboard or contribute to the design discussions and would not be able to help customers out. You can't assume that customers implies non-technical.
You'd be surprised. I've worked with some people that can say all the right things when it comes to design discussions. However, when it comes to them actually implementing the design discussions, that they themselves talked about, they struggle.
"success metric is LOC produced" == worst metric ever (even for strong programmers / IC's whose "only job" is writing code) -- and this is nearly universally accepted.
"simple technical tasks that are tedious and boring are still technical tasks. And as a senior I would never assign them to a junior developer unless there's a lesson to be learned by doing them (once)"
Given your technical seniority / superiority, whose time is better spent on tackling the thorniest challenges?
> You can't have a 5-person team where everyone is great at crunching code unless your only success metric is LOC produced
You absolutely can. Software development is a field that requires a set of intellectual abilities that are mutually (broadly, though of course individuals vary) correlated, not anticorrelated. Being great at crunching code doesn't imply being bad at any other skill required of the profession.
I can see your point, but it depends on how technical their glue work is, if you like.
In the post, she mentions for example:
> The manager has a bunch of teams and is starting to rely on Engineer to know what's going on with this one. Hey, Awesome Coder seems blocked. Do you know what the deal is?
>
> Our Engineer investigates, discovers that Awesome Coder needs information from another team but is mired in a three week long email thread. She talks to some people in real life, figures out what the confusion is, sorts it out. Awesome Coder is unblocked. He says thank you, writes thousands of lines of code.
This sounds to me like she knows exactly what's going on in the relevant part of the codebase to be able to get the information to unblock Awesome Coder.
> A customer comes in with a request: they want data that the API really should be able to provide but the team hasn't prioritised this feature yet. Our friend here spends a couple of days manually getting the data the customer needs. The customer is overjoyed.
Again, it looks like she's perfectly fine writing productive code when you let her.
I've worked with several great people in the past who exactly fit the description of "glue person" - although most of them were male - and I found that mostly, while the Awesome Coder types had an excellent understanding of their own part of the code, the glue people were the only ones who really understood how things worked together. So you got someone in team A saying why not replace libX with libY because it makes our Z easier, and the glue person speaks up and says actually I worked with someone on Team B last week, and they're using feature W of libX.
Tanya herself admits that "is this person a Senior Engineer" is controversial. But I'd say that someone who is glue in a way that clearly demonstrates a deep technical understanding of the codebase should qualify, even if that understanding is not demonstrated by writing her own code.
> Again, it looks like she's perfectly fine writing productive code when you let her.
I think this is exactly the controversial part. If the person manually gets some data and documents how to get the data, that's not writing code. That's understanding how to use a system, and can be done by an L1 or L2 tech support person.
> This sounds to me like she knows exactly what's going on in the relevant part of the codebase to be able to get the information to unblock Awesome Coder.
But yes, that's what I thought too. You often can't, it seems to me, make out these issues without deep understanding of the problem.
In general, I understand your perspective too. It's just a hard problem to solve. I would say that if I'm hired to do X and for whatever reason, I mostly do Y, that should be a huge red flag. The manager should have stepped in a long time before promotion talk comes around (as the post mentions), but it would be to me, personally, as well (did the company lie in their job interviews?).
>"I think this is exactly the controversial part. If the person manually gets some data and documents how to get the data, that's not writing code. That's understanding how to use a system, and can be done by an L1 or L2 tech support person."
Why would you stop a step short of writing code to do it for the customer? That essential time spending now to save many future hours is the root of why we code, and defines being a programmer, in my opinion.
> A customer comes in with a request: they want data that the API really should be able to provide but the team hasn't prioritised this feature yet. Our friend here spends a couple of days manually getting the data the customer needs.
This is the correct call, both technically and for the customer. The feature is planned but not prioritized, so pulling it in right now screws up the sprint. It will almost certainly also take longer to write, debug, test, and ship than it will to just pull the data - and even that assumes no UI is required for the customer to use it. A manual pull is the fastest way to satisfy the request, and that's why it's the right call from that perspective.
Note also that our friend does not only pull the data, but also documents how that was done. This means that when the relevant feature ticket does come into a sprint, it comes with detailed information and a worked example on exactly how it needs to do what it does. That'll save whoever picks it up a ton of discovery time, and that's why it's the right call from that perspective.
"Programmer" isn't the job title. "Engineer" is. There's a lot of reasons why. This kind of thing is one of them.
That's very insightful.... I think the term Engineer is used rather loosely when applied to programmers... now I get the difference. Writing the code is the easy part, knowing what to write is the skill, thus Engineer.
I think we're mostly in agreement - definitely a hard problem - and that we'd both take "hire an actual project manager and let the glue person spend her time coding instead" as a solution.
Hiring is hard for many reasons. Even on relatively short contract gigs (let alone LT or FT / roles), IME what a given org thinks it needs and what's actually required are often revealed to be quite different. To me it's not necessarily a red flag, rather a reflection of the complexity and uncertainties inherent in planning software projects. My consulting process incorporates an explicit "discovery" phase, largely for this reason.
Based on the talk [0] where the OP suggested it number of times. She mentions the engineer being rusty and having significantly fewer code contributions than her peers.
In general, if the glue person has intermediate knowledge of the codebase, they'd get promoted without much controversy. In such a case, it should not be difficult for them to bang out a couple of features and ship it, for example.
I like the term glue, I am a senior tester but a lot of my work is around just that- gluing things together, although sometimes it is more about un-sticking things.
I enjoy the large variety in my day to day work, it helps me build a large network of people and until now it made me a more essential and valued employee
I'm more of a coder (and highly prefer adhoc/throwaway POC coding) but slipped into the same thing. By being someone who fixes the annoying problems I've bumped into useful people in various departments which helps bring things together when I need to.
Of course, they now have processes to avoid favours and such so we all have to go through the same internal ticketing and scheduling system and it's completely random whether you will get someone useful and helpful or not. You don't even see a name on the reply most times.
I knew an older guy who retired a few years ago and he also called himself glue at the time. He'd long abandoned coding (he secretly admitted to me that he could do a bit, but publicly claimed ignorance) but had a good handle on what the company could do and who was doing it, which helped him talk to the right people to design feasible solutions for customers.
First, I don't get why diversity is given passing mention when there's nothing related specific to diversity or gender politics in the solid idea discussed by the article.
I also tend to value this sort of work and I aspire to be this sort of person, but it's easy to see how a company would end up incentivizing their engineers to create more stuff.
Creating stuff gives glue types their chance to optimize the small stuff.
Meanwhile, it's genuinely harder to find someone who can rise to the occasion of high technical output after doing a good glue job, than it is someone who does great technical work and can also be this sort of glue person.
Someone who does a lot of high impact technical work can identify bottlenecks and could very well do the same work as the glue person. The opposite isn't always true.
It's also much harder to figure out if someone is just very nice and doing work that would be picked up by someone else in their absence or if they are indeed identifying pain points no one else could.
> First, I don't get why diversity is given passing mention when there's nothing related specific to diversity or gender politics in the solid idea discussed by the article.
It links to the paper describing how non-promotable work is divided between genders in a mixed setting. This is clearly a diversity problem. If women take the bulk of work that won't get them promoted, you end up with women underrepresented in senior positions. Did you read the article?
In my reading of the article, the diversity point is that in a mixed-gender team, women are much more likely to be the ones finding themselves in glue roles, and this could be one of many components in the problem of why so many women leave tech (the "leaky pipeline").
The article's argument, as I read it, is that Glue Person is as good a coder as the next person when she has a whole block of time free in her calendar, but the problem is that her calendar is full of meetings.
One thing I like about this post is that it presents an actionable way to do something to improve diversity: managers should check if the women in your team are disproportionately loaded with the kind of task that's essential for the project to succeed, but doesn't count as much for promotion metrics.
> First, I don't get why diversity is given passing mention when there's nothing related specific to diversity or gender politics in the solid idea discussed by the article.
If you can get promoted for many things in multiple ways, that means diverse promotion system, which means you'll get more diverse seniors and more diverse leadership.
If your goal is to have a wide range of skilled people from various backgrounds in senior places, relaxing rules around promotions is definitely a way to get closer to that goal.
If you have a strict promotion criteria, you might end up promoting very similar people.
An extreme example: Only people from Ivy League schools can get to senior and above.
Consequence: you'll end up with very heterogeneous leadership 5-10 years of such a policy (most likely even leadership that formed friendships while in university).
> You'd like to have time to code, but nobody else is onboarding the junior engineers, updating the roadmap, talking to the users, noticing the things that got dropped, asking questions on design documents, and making sure that everyone's going roughly in the same direction... If this describes you, congratulations: you're the glue.
Edit: Looks like the title was changed to "Being Glue" which is probably a better title.