Platforms change but cool URIs don't. @ Irrational Exuberance

Hi folks,

This is the weekly digest for my blog, Irrational Exuberance. Reach out with thoughts on Twitter at @lethain, or reply to this email.

Posts from this week:

- Platforms change but cool URIs don't.
- Interim assignments.
- Stuff I haven't written yet, but hopefully will someday (2022).

Platforms change but cool URIs don't.

With the recent news of Twitter’s board accepting Elon Musk’s offer to buy Twitter, some folks are talking about leaving Twitter. In the long scheme of things, being founded in 2006 makes Twitter a young company, but the internet is different and over the past 16 years it’s become a central platform for many folks working in the technology field (among many others). Twitter has become especially important for folks writing content online, to the extent that it’s the most effective distribution mechanism for many writers. This is certainly true for self-published authors doing their own marketing, but I’ve also heard stories of publishers stipulating in their contracts that authors must be active on Twitter as well as partially filtering prospective authors by the size of their Twitter following.

I hope that Twitter remains a vibrant, effective forum for discussion as well as an effective distribution channel, but it’s an important reminder that even the best platforms don’t last forever. Tumblr and Flickr were the home of many vibrant communities before they were both acquired by Yahoo! MySpace similarly faded out after being acquired. Digg’s tech platform and data didn’t survive its acquisition. Facebook is still around, but the early APIs that powered its application ecosystem are long gone. Even Twitter itself shifted from encouraging a rich, API-centric, developer ecosystem to implicitly shutting down most API usage.

You can certainly argue that those platforms aren’t really content or distribution platforms, but it’s hard to look at recent subscription newsletter darling, Substack, without thinking about the increasingly unpredictable paywalls of yesteryear’s blogging darling, Medium. In theory you can simply replatform every five or six years, but cool URIs don’t change and replatforming significantly harms content discovery and distribution.

For folks who invest a great deal of time into creating content online, I think this ultimately means that you need to own your content, own your DNS records that connect readers to your content, and own your mailing list. As long as you control those interfaces, you can move platforms without impacting discoverability. Sure, it’s a bit of work, but this blog has moved five times, from Dreamhost to Linode to AWS to GCP to Github Pages, over the past 15 years and as far as I am aware all the URIs still work. If I’d routed folks directly to Medium or to a Github page on the Github domain, I wouldn’t have been able to do those migrations cleanly (incidentally those platforms didn’t exist when I started writing, but that’s not really the point here).

Probably the most important URI is the one where your RSS feed lives. RSS is a much less important distribution mechanism than it was a decade ago, but RSS readers are some of the most valuable readers because they include the subset of folks who first discover content to reshare on various social news platforms.

For a long time I held fast to the dream that RSS would be enough, but I eventually started a mailing list as I started to ramp towards releasing An Elegant Puzzle. Even today that list isn’t particularly large, but it’s still over 6,000 folks that I can reach directly even if Twitter disappears tomorrow (Staff Engineer’s mailing list is a bit larger, Infrastructure Engineer’s list is brand new and very small). It’s also a highly engaged subset who are much more likely to reply to the email with their thoughts. I’ve reluctantly come to believe that URIs and email are the durable interface and protocol that will live long past every given platform’s peak adoption (although, just imagine the sheer chaos of gmail shutting down one day, I do regret that I haven’t yet moved my email to my own domain, probably a good project for this year).

If you plan to write across decades, you simply must own the interfaces to your content. You should absolutely delve into other platforms as they come and go–they often become extraordinary communities with great distribution–but they’ll never be a durable home.

Interim assignments.

One of my favorite parts of senior leadership roles is that you periodically get to deep dive on something that you typically don’t focus on too closely. At Stripe, I got to serve as the interim leader of the Payments Infrastructure organization for a few months, which gave me the chance to support a couple hundred additional folks working in an adjacent area of the company, while also continuing to support the Foundation organization I worked with throughout my time there. I learned a disproportionate amount in that window, almost as if I’d started a new job, but without needing to build a whole new set of relationships.

More recently at Calm, our data science team has reported through me since I joined in 2020, but we always had a data science leader who was directly supporting the team and refining our approach to data science. Last year our data science leader stepped down after a long run at the company, which gave both me and another Calm engineering leader the opportunity to get more deeply connected with the team while we hired the data science leader for the next phase of our growth.

Interim assignments to support an area while its long-term leader is being hired are the manager’s version of an individual contributor “doing a rotation” onto a different team for three to six months to support an important project. Whereas individual contributors often get to temporarily switch teams, I’ve never seen it work that way for managers. Instead managers end up managing two teams because they wouldn’t be hiring for the role if leadership believed there was internal capacity to take it on permanently. In your three to six months of the interim assignment, you’ll probably have a small, but likely meaningful, impact on the team you support, but the context you take away from the experience will significantly increase your impact once you return to focusing on your ongoing duties.

While I’ve never seen a manager directly rewarded for taking on this sort of endeavor, they’re a valuable leadership experience, and I’ve often seen them indirectly lead to larger roles by proving out your, and your team’s, readiness. When you’re leading a well-run organization, you’ll often find calm pockets of time where you’re a bit bored but also know that it’s only calm until the next thing goes awry; interim assignments are a great way to take advantage of those pockets without compromising your workload if things get busy in a quarter or two.

The second most valuable thing about interim assignments is that you don’t have to take them. If your current area, or your personal life, is going through a bit of chaos, then don’t take one on! If things are feeling a bit too easy, then let your manager know you’re interested and available. Whereas growing your permanent scope in a larger organization can be a tricky process with many stakeholders, interim leadership is usually quite straightforward to achieve: folks compete much harder for ownership than they do for work.

Stuff I haven't written yet, but hopefully will someday (2022).

I took some time and cleaned up my writing backlog of “topics to write about” and pulling them into this post. Let me know if any topics sound particularly helpful, or if I have any particularly good stories that I’ve forgotten to mention here.

Stories recounting interesting times of my career or life, although inevitably the very best stories are tricky to talk about in public, like these previous stories:

• “From ten to 2,000 services: the Uber provisioning.” Lightly touched on in Trunk and Branches
• “Starting Uber’s SRE organization.”
• “How increment.com was founded.”
• “Uber’s 18 hour PostgreSQL outage.” Undoubtedly the most chaotic outage I’ve been part of
• “The writing stopped.” Why did I write five blog posts from December, 2012 to May, 2016? Compared to 12 from August 2012 to December 2012 and 14 from May 2016 to December 2016

Topics in my writing backlog:

• “It depends: effective approaches are fact-specific.” A lot of leaders, if you chat with them, implicitly don’t believe that the quality of their decisions matters. My experience is that the quality of your decisions do matter a lot, and that it’s pretty easy to make reasonably good decisions if you put effort into it and use simple tools
• “Writers who operate.” I think the most interesting writers are operators, e.g. folks who write about the actual work they do from the perspective of doing that work. Without the constraints of operation, it’s easy to get too overly pure or hypothetical in approach
• “Hold on to your hats.” 2022 is already a wild year to be working in the tech sector. It’s probably going to get a lot messier, although, tbh, it might all recover without much prolonged consequence. How do you operate in these circumstances?
• “100 simple decisions quickly and 5 hard ones: a quarter of executive work”
• “Recruiter alignment/prioritization should happen wherever headcount planning happens.” Too many organizations try to set recruiter priorities at a different node in the organization than where the organization they support sets hiring priorities. This doesn’t work very well. Related to the piece on Headcount Planning I wrote
• “Avoid organizational sink holes.” Basically, what are other tactics when you’re caught in a “hard to work with” trap
• “Avoid people who are stuck on a single idea.” And relatedly, “Avoid novelty driven development”
• “Intuitive architecture doesn’t scale.” The intuitive approach to architecture works well in small companies and small, isolated teams, but it doesn’t work in larger organizations with cross-team dependencies. If you want to be a senior technical leader in a scaling organization or a large company, you have to do structure architecture rather than intuitive architecture
• “Determining engineering compensation bands.” There’s a lot here, especially in 2022
• “Great engineering leaders tolerate uncertainty andd espite misalignment.”
• “Right people before right process.” Good process doesn’t always generate good decisions. Excellent organizations prioritize good decisions over process
• “Learning with fewer mistakes.” This idea reminds me a lot of a story my mother told me about starting out nursing at a teaching hospital: it was just assumed that new doctors would kill a few folks while they figured it all out. Yikes! Management is generally less lethal, but still a lot of room for harm, and even today I find that much of what I learn comes from understanding my mistakes. How can we learn as managers while minimizing the negative impact on folks around us?
• “Be better in five years than in six months.” Ramping slowing but durably into a new job, career, etc
• “Mirage metrics and headcount planning.” A lot of headcount planning is anchored on made up, useless metrics like engineering velocity. Probably most of these don’t matter, and are just a proxy for how much leadership trusts you
• “Match process with organizational complexity.” The biggest cause of failed process rollout is trying to pull in something that worked at your old, much more organizationally complex company
• “How do you measure your impact on the industry?” I don’t actually know! But it would be interesting to have an answer here
• “Hypergrowth playbook versus bankruptcy playbook: do you have high efficiency user acquisition vehicles to drive spend into?”
• “How should you do engineering diligence in acquisitoins?”
• “How can your organizaton be inhospitable to sealions?”
• “It’s OK to make some bad decisions.” Work decisions across rounds not within rounds. Learning from bad decisions and learning overall is usually more important and less disruptive than trying to reverse somewhat bad inflight decisions
• “The waiting place.” A Dr Seuss book, yes, but also a common experience for startup employees with illiquid equity
• “What are manager archetypes?” I actually have a draft of this Engineering Manager Archetypes that I’ve never published because I don’t think it’s particularly useful
• “Why does every company hate their knowledge base?”
• “Be a change sink.” aka absorb friction from change in your organization
• “The ‘If it isn’t as good as I can do it, it’s bad’ variety of gatekeeping.”
• “What are some real-life ratios of staff-plus engineers to director-plus managers?”
• “Are you a company that specifically facilitates first-time entry to Staff roles? How and why?” Plus an exploration of intentional and accidental title inflation
• “What are career defining roles?” I think this is an interesting topic, basically, why are some roles specifically career creating for the individuals who have them?
• “Sabbaticals: do they work?”
• “The flying wedge.” Why do some groups of folks join companies in a sizable group? How does this dynamic impact the receiving company? Is this a good or a bad thing?
• “Using nudges to drive migrations.”
• “The emperor’s unicorn cothes – conviction nd honest when everyone is afraid of losing generational wealth working at their unicorn startup.”
• “Good career ladders reflect real cultural values rather than anspirational cultural values.”

That's all for now! Hope to hear your thoughts on Twitter at @lethain!

This email was sent to you why did I get this?    unsubscribe from this list    update subscription preferences Will Larson · 77 Geary St · co Calm 3rd Floor · San Francisco, CA 94108-5723 · USA