The Tiny Teams Playbook | Hacker News Reader

The Tiny Teams Playbook

latent.space

137 pointstilt46 comments

The office or otherwise mandatory frequent in person work sessions bit seems pretty at odds with the underlying idea that you’re a team focused on actually delivering and building with deep focus. What does commuting a half hour, hour, or more, each way to an office to put my headphones on and zone in, do to achieve any of that? I’m gonna be able to do that more effectively, more focusedly, and at the hours I’m most productive, remotely. The commute is strictly a distraction.

Same. I find myself much more productive. I do like coming in every once in awhile for the rapport and cultivating working relationships face to face though.

Yep, 2-4x onsites together a year to develop human relationships, and otherwise 100% remote, is by far the most effective team arrangement IME. It is especially the most accessible format for people who do not necessarily perform their best work within the typical office hours (9-5, +/-1), who do not want to live in your metro area, or who are distracted by disturbances in their surroundings - or just aren’t hardcore extroverts.

Or simply put: if you truly want the best, most focused, highly performing team, an office requirement shrinks your talent pool tremendously for extremely little gain. Do quarterly meetups somewhere and move on, IMO.

Quarterly is a lot, depending on travel distance and whether weekends are needed for travel, for folks with families.

Agree, 4x is doable only if you have no kids. Otherwise you not only create a lot of stress for the other parent but you also lose valuable time with your family, which you won’t recover as kids grow up. I feel like 2x is reasonable.

I don't have kids and I'm still more on team 2x/yr than 4x/yr by a large margin, but I'll still take 4x/yr for ~3 days each over in-office daily (or even twice a week). Mentioned it mostly as a compromise option I could accept (and IIRC that compromise is what 37signals did and documented in the Remote book which I haven't read in quite some time).

The commute doesn’t help you, but working in an office next to your team mates will accelerate your work.

Software development is a team sport and individual productivity is not the same as team productivity. Communication bandwidth in person is much higher when colocated. Startups move fast and higher bandwidth increases velocity, reduces errors, improves quality and team cohesion.

For other situations remote can be “good enough”, and has advantages eg bigger recruitment pool or cheaper labour, but in general in person is just going to be a lot faster with higher quality results.

A lot of engineers don’t wish this to be true, because wfh is often better for them as individuals, but it is what it is.

I’ve worked in plenty of startups (the overwhelming majority of my career, actually) and did not perceive the performance of in-office teams to be significantly better than the remote teams I’ve been on. The floor is probably lower for remote teams (in that ineffective remote teams are horribly ineffective), but the ceiling is comparable, and the average is (again, in my experience) anywhere from comparable to slightly better, because folks are working the ways+hours they’re most effective, not what someone else thinks should be the most effective.

I think it depends on your job role. I’m more architecture and operations in past lives, and being together is really powerful and reduces time taken for many tasks.

If you’re an engineer or developer mostly working a backlog, totally different story - wherever you are most comfortable working is ideal.

Either way, dogma is terrible. I have a friend who is a specialist in a specific area of finance who has been WFH for 20 years. Now she’s commuting to an office in a city about 300 miles away from the rest of her team, because the big boss says come to the office.

> The floor is probably lower for remote teams (in that ineffective remote teams are horribly ineffective), but the ceiling is comparable, and the average is

Agree, and the difference between the floor and ceiling is typically leadership. Not just by the executives but everyone who works there.

I use this kind of opinion as my idiot bat signal now. It’s so obviously untrue when someone starts spouting this nonsense you know they are a very feelings based decision maker.

I have been leaning the other way. There’s room for nuance in the discussion but a stance of certainty that full remote is just more effective screams “expert beginner.”

There’s a third option: some people work best alone / remote. Some people work best in an office.

There’s more than 3 options. That’s what the nuance is about.

Sometimes individual productivity isn’t even the measure of success and sometimes it needs to be sacrificed for group productivity.

> some people work best alone / remote. Some people work best in an office.

this is true, and irrelevant to my reasoning.

Got any evidence of this or is just vibes based?

Unsure why the status quo needs evidence but remote doesn't, but which part of my reasoning do you require evidence to believe?

The two most productive startup software teams I've worked on have been 100% remote. The least productive was 100% colocated. Anecdotal evidence only but so far I'm not seeing any reliable quantitative evidence for your claim.

"software development is a team sport"

I have heard people state this like it is obviously true. Not so sure.

> The office or otherwise mandatory frequent in person work sessions bit seems pretty at odds with the underlying idea that you’re a team focused on actually delivering and building with deep focus.

It’s at odds with being pro-AI too. If you can’t effectively collaborate with somebody who is not physically present but is just a voice on a call or words on a screen, what does that imply about how effectively you can collaborate with AI? Are these people building robot bodies for Claude’s GPUs and making those robot bodies sit in office chairs across from them? Would that make them more productive?

But AIs aren't humans, so why would we need to interact with them the same way? AIs also don't fill the same role in a team as humans do (at least, not in the kind of team discussed in this post).

> But AIs aren't humans, so why would we need to interact with them the same way?

One could argue that some AIs are like SOME humans. After all, you and I are both humans but we are not the same and no reason why our colleagues would interact with us in the same way. Nuance

> Camaraderie, speed: Have fun, do retreats, avoid burnout

Or, alternatively, respect personal boundaries and don't force coworkers to have social outings.

I really wish "work is just work" was more popular. There is an empathetic way to do this that isn't just treating people as a number but also not forcing socializing outside of the context of work.

Yes to avoiding burnout. No to thinking a retreat is the answer to that.

Hacker News formula for startups: no offices, no offsites, no meetings, and no MBAs. If only idiot CEOs and rapacious VCs were listening!

Not quite.

Have an onsite team or have hybrid setups that bring people within geographic areas together. Nothing replaces getting around a physical whiteboard in a physical space.

Context is in the original statement that retreats are a fix for burnout.

I used to agree with that but now we have tools like Figma that actually work better than a whiteboard once you understand how to use them correctly.

Sure Figjam is great but my anecdata is that a majority of the value of the whiteboard is the discussion as it happens not the resulting artifact. Lucid works better for that because it let's you easily frame things into slides (yes, a section in figjam is usable for this but it's not the same).

The concept being that you break things down and layer in details so the evolution is explained similar to the train of thought from live sessions.

I've been remote almost exclusively since 2012 and have used everything out there... the biggest thing that makes an improvement with remote comms? A very high quality microphone with a high quality compressor / gate (hardware, even) and using Mumble / Teamspeak / etc with a good codec where everyone can hear everyone speaking at the same time with low latency. That's what we lost from copper phones and it's exponentially worse in Zoom where you can't tell when someone is trying to interrupt you with the auditory cues. It really makes a big difference.

I'm an extrovert so I love video / cams on but really audio is all that is needed when paired with a tool like Figjam and getting high quality audio makes a remarkable difference.

The kind of people in these small teams are not ones to think "work is just work".

(i'm the author) thanks for posting OP! here's the youtube full playlist https://www.youtube.com/watch?v=pQz-PgA1eJw&list=PLcfpQ4tk2k...

more thoughts after about 6 months of stewing with the Tiny Teams idea:

the more work experience I have in tech, the more I see the inverse relationship between size of team and velocity on projects. I think Zuck aside, the race towards 10-20mm comp packages (this is real btw) for high velocity AI engineers (both the kind that are very good at using coding agents and the kind that ship AI products) is a direct economic consequence of this very human observation meshed with the reduction in cost of shipping software as long as you have a very good supervisor/prompter/architect to keep things on rails.

I actually think the biggest casualty of this is 1) people with "bullshit jobs" in tech e.g. "product managers" that are actually "project managers" that call in on zoom from their poolside to ask "ok what's your ETA on that?" on their jira board twice a week, and 2) the VC industry since (if you dont pay a ton of cash comp) companies are close to profitable (https://www.swyx.io/cognition) after an initial ramp due to the insane labor leverage. the one-and-done round i think is going to be increasingly common in VC.

This was a very challenging article to read. Not because any of the concepts described, but for the way ideas are thrown around and organized. This looks like it was written by a set of llm agents that were instructed to write an article without a clear outlined, and then the author took what they felt were the best bits and hit publish.

(i'm the author) it was 100% human, i'm afraid to say. i was trying to summarize a lot of content in a concise article.

An erroneous summarization is:

  The Tiny Teams concept has resonated so strongly that it’s 
  pretty clear it is the next major transition of the org 
  chart as we go from level 2 to 3 AGI.

LLM's, nor any offerings based upon them, qualify as Artificial General Intelligence[0]. So to assert there is an existing "level 2" AGI, let alone a progression to "level 3" AGI, is nonsensical.

0 - https://en.wikipedia.org/wiki/Artificial_general_intelligenc...

Quoting wikipedia:

>Researchers generally hold that a system is required to do all of the following to be regarded as an AGI:[34]

> - reason, use strategy, solve puzzles, and make judgments under uncertainty > - represent knowledge, including common sense knowledge > - plan > - learn > - communicate in natural language > - if necessary, integrate these skills in completion of any given goal

Modern AI can do all of those.

Can it though? Everything I've seen and experienced is that LLMs are very good at making it appear to do those things, but the amount of times I've gotten stuck on "you're absolutely right!" When correcting the LLMs suggests that it can not reason by any means, nor does it learn. Otherwise, an LLM would never get stuck in a loop.

Of the reproduced attributes listed (reformatted and numbered):

  1 - reason, use strategy, solve puzzles, and make judgments under uncertainty 
  2 - represent knowledge, including common sense knowledge 
  3 - plan 
  4 - learn 
  5 - communicate in natural language 
  6 - if necessary, integrate these skills in completion of any given goal

Only 2 (partially) and 5 are applicable to LLM's, as they are statistical text (token) generator algorithms. Very useful ones, no doubt, but not ones which satisfy all of the above.

AI is mostly trained on average stuff and the author is most widely known for throwing a bunch of stuff on the wall at different communities, they were never known for quality.

Also didn't care for the hiring section and how dystopian that feels, well unless you're independently wealthy I guess. People really want to recreate monarchy in the work place.

"Simple, Boring Tech Stack:" Good advice but bad example, because it depends on what engineers are familiar and comfortable with and technology itself should be mature enough. You dont want to spend time building orchestrator when k8s solves it for you. Most cloud provides provide you with k8s as a service, which are miles better than using shell scripts, if you are already familiar with k8s

"Almost no meetings: “deep focus” - building instead of talking about building"

I used to work in an environment with often 8 hours of meetings straight. People had their headsets on while being in meetings and were simultaneously programming and when they heard their name mentioned they tried to say something smart. It was a terribly inefficient way to work.

Then I switched to an environment where we took "Almost no meetings" seriously and it was a tremendous boost. After a year or so I realized that we left a lot of potential efficiency untapped because of lack of communication or miss-communication.

Now I think there must be a middle ground - an optimum of communication for an optimum of efficiency. Teams need to be actively steered to that, just hiring good communicators and hoping for the best is probably not going to work. You need meetings. At least some. And some seemingly inefficient meetings will prevent inefficiency elsewhere.

Everything I wrote above was about highly distributed teams working remotely. The Tiny Teams Playbook has also

"In Person: either have an office, or VERY frequent AirBnB hack weeks"

in it, which changes things quite a bit.

That middle ground for me is what I like to call "proposal driven development".

Ideas, concepts, implementation plans are first written down as a proposal, which is read by others and discussed online. Meetings are only required if there are blockers to resolve, or differences in opinion.

`Simple, Boring Tech Stack: shell scripts over k8s, keep code modular`. wut

I think now k8s is boring and shell scripts exciting.

focus on the customer's problem. Not on debugging a huge thorny k8s stack or AWS deploy.

> Top of market salaries: 95th+ percentile salaries

Absolutely. Pay, in real money, today. (as opposed to promises of future riches if all goes well)

If you're trying to do more with less, you can't afford not to.

This is the sort of inconvenient detail that a CEO will omit when implementing the others (like onsite) and claiming "yeah we're fast moving we follow the Tiny Teams Playbook"

Reply on news.ycombinator.com