Robot Girl Gang

The Latest Musings

  • Commodification: The Race To Zero

    Steve Crowe of The Robot Report has been following Teradyne’s adventures in IP protection this year, most recently noting their second foray into court against the Chinese cobot manufacturer JAKA. Our bickering over parking and their absolute domination of my colleagues in the neighborhood basketball tournament aside, I’ve got mad respect for them. Teradyne has been a force in the automation industry: they make some of the best automated electronics test products in the world, and their Universal Robots line appears regularly in our own workcells. So naturally, when I see Steve covering Teradyne, I lean over the fence for the neighborhood gossip. But when he asks, “is this simply what happens when an industry matures?” it finally knocked loose some observations that have been rolling around in my own head for some time.

    In a word: yes. This is a sign of industry maturity but for a more dreadful reason: it’s a warning sign of commoditization.

    Let me step us back for a moment and talk about some finance bro stuff. You may have heard finance bros and VCs talking almost fetishistically about “alpha”, without ever mentioning “beta”. Here’s some quick definitions:

    • Beta: Your commodity baseline, what any competent manufacturer can produce. A 6-axis cobot is as a 6-axis cobot does. Baseball terms would put this as your replacement-level player: one who can be called up that performs at the minimum standard for minimum cost.
    • Alpha: Your excess value above baseline, the stuff that makes your product stand out and seem more attractive than that Beta baseline. Could be the integration, the ecosystem, the system design, the deployment experience, the support quality… anything that justifies your charging a premium for your product over the functionally equivalent alternative. This is a baseball player with a positive WAR[1].

    Alpha constantly erodes due to technology and manufacturing advances, greater knowledge in the market about how to solve problems, rising customer expectations, and other pressures. Last year’s alpha becomes this year’s beta. Bluetooth was once a premium feature, high alpha; now it’s table stakes, strongly beta. This natural erosion is called “alpha compression”, and it’s a race to zero as the alpha continually dissolves away towards nothing.

    To bring it back, commoditization occurs when your alpha evaporates. That unique thing that you brought to the market isn’t so special any more.

    Now, commoditization is not a universal evil. We need commodities, both virtual and physical. The entire Open Source Software movement is a response to the need for commoditized software. If I needed to find a “special” machine screw for every application I would lose my mind.

    But commoditization requires a radical shift in your operating model: driving towards massive scale, absolutely relentless cost efficiency, and volume is your only lever. You can do that in high tech: Kingston’s our reigning champion of massive scale and reliability in the memory module market. However, when you lose your alpha before you can transition to your commoditized operating model you’ll get ground under by competitors who can race to the bottom faster than you, just like the photovoltaic industry did[2].

    So how do we diagnose commoditization, and how do we grade it?

    First off, we need to recognize that companies all use market barriers to protect themselves. Some barriers, such as intellectual property instruments like patents and copyrights, can be virtuous: governments recognize that reasonably-sized paths towards extended alpha stimulate investments in creativity that benefit society as a whole. The problems happen when companies use these barriers to the exclusion of other alpha-creating activities such as R&D investment[3].

    Commoditization has three major stages.

    • Early stage: Ecosystem lock-ins as substitutes for genuine product improvement. John Deere’s war against Right-To-Repair provides a perfect example of weaponizing the pain of switching to a new vendor to guarantee market exclusivity. John Deere still has plenty of alpha to exploit but they’ve instead chosen the lazy path that doesn’t require creativity or risk.
    • Mid stage: Regulatory capture, standards manipulation, all that devious stuff that effectively writes out anyone else from being allowed to compete in the first place. AT&T has committed many sins, but the best example of mid-stage commoditization was AT&T’s history of abusing their monopoly position to prevent third party devices from being attached to their phone lines. That meant no answering machines, fax machines, dial-up modems, or assistive devices not produced and sold by AT&T themselves. Once the FCC finally lifted this restriction in 1968, telecommunications devices exploded and AT&T sold more phone lines than ever per capita, going from 414 lines per 1,000 people in 1960 to 796 lines per 1,000 people in 1980[4]. AT&T played themselves by trying to control alpha when they should have focused on their beta.
    • Late stage: The final argument of every petty asshole with nothing else left to offer: “I’ll sue you!” It’s ugly, it’s expensive, it’s messily public. Sometimes you have to go there; for example, trademarks require aggressive protective measures in order to maintain them, such as Patagonia’s uncomfortable infringement lawsuit against Pattie Gonia[5]. On the other end of the spectrum you find patent trolls like SCO, who never had alpha and honestly have no beta left either. Perhaps they only have rho. Most IP protection lawsuits land somewhere in between, like Teradyne’s, where there’s legitimate beef on patent infringement but that’s the only thing they’ve got left in the competition space.

    So using that guide, let’s track the progression of the disease for Teradyne’s Universal Robots. In The Beginning, UR created the Cobot, and saw that it was good. PolyScope kicks ass, arms are stupid easy to replace in the field, real safety co-existence. As time goes on, switching to other companies’ products generates its own kind of lock-in and collaborative safety standards continue to provide a considerable buffer. Then, Chinese manufacturers manage to close the hardware gap, not only collapsing UR’s alpha but compressing their beta as well. Teradyne’s now suddenly in a fight for their lives because the hardware should never have been their wooden walls safeguarding their city[6].

    Commoditization comes for all markets, and one that’s probably most tactile on an everyday basis is the personal computing market. Most PC makers are optimizing their beta: Dell, HP, Lenovo. They assemble laptops and desktops from roughly interchangeable components that fade into the background compared to the price and spec sheet and whether you can remove the stupid “register your Windows license” watermark. My personal spite for modern HP products aside[7], you won’t generally find a user who ride-or-dies their Dell Latitude. The Fruit Company on the other hand, can only survive by maintaining their alpha. Sure, Apple maintains a hefty IP portfolio but they earn their market loyalty through tight component integration and great software-hardware co-design, even in the Intel Mac days when they were running the same processors as their beta compatriots. When they rolled out Apple Silicon, they then added hardware differentiation back on top of that already fat alpha.

    Coulda, woulda, shoulda. Teradyne’s now in a bind. They might still be able to dig it out and find alpha in PolyScope’s UX, in creating an App Store equivalent for robots with the UR+ ecosystem, or in building an authorized reseller army from their integrator network. Can they turn it around like Apple did in the 2000s, or do they fight for pennies like a Dell or Lenovo?

    It’s entirely possible, and even feasible. Teradyne in general and Universal Robots specifically have great people (even though they slow to 3mph for 15mph speed bumps, wtf). It’ll take creativity, a little bit of risk taking, and a willingness to experiment. No need for a moonshot here, just some cleverness and the space to implement it. But IP protection, including their lawsuit, doesn’t buy you alpha, it just buys you time. The race to zero won’t stop for a court date.

    [1] – Wins Above Replacement. Basically, how much more the player contributed to winning a game than your baseline player would. The Sabermetrics people get way too nerdy about this way too quickly, but Wikipedia’s got a decent explainer for us mere mortals.
    [2] – No one’s pockets are as deep as companies backed by state-sponsored capitalism.
    [3] – Pharmaceuticals are an entire problem with this unto themselves.
    [4] – Munged together from US Census data and the FCC’s Statistics of Communications Common Carriers.
    [5] – Suing her for $1, because that’s the minimum to demonstrate they’re protecting their trademark.
    [6] – Oh I’m sorry, we’re going to use Greek letters and I’m not going to slip some classical history in? Pulled a fast one on you.
    [7] – The last great HP product was the LaserJet 4 with the Jetdirect Print Server card and I will not be accepting questions. That absolute unit lasted for two decades after I tactically acquired it.

  • Why They’re Gross

    Likely my second-most career-limiting belief, right behind refusing to make the tools of war. This is the spicy one where I’m planting my flag because it goes to the very heart of how we treat and regard each other as human beings[1]. And while I’m not gonna die on this hill, somebody is.

    Let’s first define a humanoid robot. A robot has its functional safety system, the physical hardware, the controls software that commands its motions, and the decision-making software that chooses what functions it performs when. Placing these components in an anthropoid form factor with a head, torso, arms, and legs, makes it a humanoid robot[2].

    We’re surrounded by robots in our daily lives[3]. So what makes humanoids so icky?

    I’ve been loudly anti-humanoid and anti-pedalist[4] for over a decade, but it’s been from a functionality-first Technical Product perspective. Sure, backflips are cool! But, unless you’re a Savannah Banana, they don’t really accomplish much. I never got the “so what?”, and didn’t see them driving much in foundational technology progress such as energy management, servo development, or motion planning. They’re cool primarily because they’re built in the form of their maker, and we can imagine them as generalists because we are generalists ourselves. But that’s a technical answer which doesn’t address that ick factor underneath.

    Humanoids are a white-hot market full of solutions in search of problems. And yet, we’ve had an explosion of humanoid robots soaking up VC money and sometimes even hitting the market despite the absence of a specific problem to solve. They’ve been developed as technology-first initiatives with the hopes that a killer use case will reveal itself somewhere down the line. There’s no purpose, no vision, and no obvious path for how they’ll improve our lives.

    In the absence of that vision of improvement, the least imaginative amoral profiteers among us have happily stepped in to fill the gap. These wanna-be neo-feudalists[5] have no desire to improve processes or reimagine how work can be done better, faster, and with a lighter footprint. They just want cheaper labor doing exactly the same thing as before, just without all of that pesky complaining or lunch breaks.

    When you drive labor costs and labor rights to zero, you get chattel slavery.

    It’s the next song on the same album with such bangers as “Chatbot All The Things”, “A Spreadsheet Wrote This Article”, and “Literally All The Russian Psyop Videos”. Quarterly results-driven decision-makers have accepted the tradeoff of lower quality to reduce the number of human beings they employ without thinking through how these tools could instead truly transform their work. A quality product no longer enters the conversation, only the fastest path to your wallet. And both we and they are worse for it.

    The profiteers, the chattel slavers, and the neo-feudalists all assume a zero-sum game, and thus are racing to the bottom. Zero-sum games end in industrial commoditization and social dystopia because the players cannot imagine creating something larger than what they’re currently fighting over.

    THAT is why properly educated[6] citizens of free societies get the ickies from humanoid robots: unlike other robots, we sense that they have no other purpose than to devalue and commodify us. And our industry’s incredibly thoughtful and nuanced response to this existential threat is to “make it look less creepy”. This form factor, without function, is an insult.

    Great automation augments human labor rather than replacing it. We’re social animals by nature and will pack-bond with anything— anything[7]. We like making friends with stuff that joins the team alongside us and doesn’t supplant us.

    Humanoid robots can’t earn their place as part of our team until we figure out what they’re uniquely good at compared to a human being. It’s why you should immediately look upon any tele-operation training initiative with great suspicion: they’re not offering us anything new.

    So what are they good for? Humanoid robots are my bet for when we need somewhat bespoke behavior in somewhat bespoke tasks performing human-shaped work in situations that we shouldn’t be sending our fellow human beings. Let’s take that apart:

    • Somewhat bespoke behavior: a well-defined task that requires variation, but not innovation, within some well-defined boundaries. Picking up an object they’ve never seen before is a great example.
    • Somewhat bespoke tasks: Actions you won’t be performing frequently. If it’s the same rote task over and over again, you need an industrial engineer rather than a humanoid robot.
    • Human-shaped work: Here’s the dangerous one because it’s so easy to be lazy about it. Specifically human-shaped work requires a human to interface with it and it can’t be redesigned in time to accommodate other form factors. Emergency responses in legacy industrial plants with valves designed for human hands, for example. Packing warehouse orders, on the other hand, ain’t it, chief.
    • Situations that we shouldn’t be sending our fellow human beings into: If you need a specially trained human being to suit up to go in, there’s your answer. Burning buildings, confined spaces, extreme environments, nuke/bio/chem hazards, you’ve got the idea. It’s not a permission slip to design dangerous workplaces, it’s to address scenarios where you cannot design out the hazardous exposure.

    By analogy in the non-humanoid world, bomb defusers were some of the first robots to gain wide workplace acceptance and affection because they could fulfill most of these criteria. Exactly zero EOD specialists felt dehumanized by them.

    So if you truly want to make humanoid robots, do it right and start with your intention explicitly. “I want VCs to pay me to make humanoid robots do, um, stuff” invites the least principled in among us to take advantage of your work to profit from your fellow human beings in whatever way they can get away with.

    Be a human enabler, not a slaver.

    [1] – Look, this post is tagged as “Philosophy” for a reason. Buckle up, buttercup.
    [2] – No, Brad, that’s not a humanoid. You’ve overshot clickbait straight into cringe.
    [3] – See A Dishwasher Is A Robot.
    [4] – I’ve got a whole rant about the utility of the human foot and it is 100% not a weird Quentin Tarantino-type thing.
    [5] – If this characterization offends you, kindly go fuck yourself. You are, in fact, the baddies.
    [6] – Ignore the parchments on the wall; auto-didacts are welcome in this home. Classically speaking, a person who has cultivated the cognitive skills and tools to participate in a liberal democracy. Curiously, liberal arts are the disciplines most denigrated by neo-feudalists who deem them dangerous to their objectives.
    [7] – Once again, see the ever-growing body of work on “Captain Stabby”.

  • A Dishwasher Is A Robot

    One of my favorite ways to troll roboticists is to claim, very loudly, that “a dishwasher is a robot”. It’s a great litmus test for how a given person will see the industry because it uncovers an uncomfortable taxonomic void that we haven’t yet resolved as a people.

    Let me start by explaining how a dishwasher is, by many definitions, a robot. ISO 8373 tells us that a robot is generally a “programmed actuated mechanism with a degree of autonomy to perform locomotion, manipulation, or positioning”. Well, looking at my dishwasher at home, we’ve got:

    • Programmed – I can select various modes that I’d like the dishwasher to run in. I have cats and a husband, so my dishwasher generally stays in “heavy / extra hot”. My dad’s has a cycle gentle enough to run Grandma Mae’s gilded china in.
    • Actuated mechanism(s) – Water pumps, heating elements, fans, sprayers, there’s lots of actuated components here.
    • Degree of autonomy – I can set it and forget it. My dishwasher can operate without me and make decisions to heat water, to what temperature, when to release soap, when the dishes are sufficiently rinsed, and so on.
    • Performs manipulation – The systems out there these days are amazing. Complex multi-jet arms, spinning discs, adaptive sprayers, smart soap release mechanisms… it’s a whole thing.

    Add on the subclass of a service robot, defined as performing useful tasks for humans or equipment in personal or professional use (eg, “cleaning”), and we’ve nailed it.

    So why don’t we think of our dishwashers this way? Or our refrigerators, laundry machines, or an entire building?

    Because they became so familiar that we only notice them when they’re missing. Dishwashers ceased being seen as robots the moment people stopped calling over their neighbors to marvel at the new technological doodad they purchased and started getting annoyed when their apartments didn’t include them. It’s the best destiny of any technology: to become invisible, reliable infrastructure. Robots become appliances the moment they achieve ubiquity.

    And here’s how you know that a technology has achieved ubiquity: we build around it. Kitchen designs accommodate fridges, stoves, dishwashers, and now microwaves with dedicated spaces and utilities. Your home has heating and cooling infrastructure threaded throughout in the forms of sensors, ducts, compressors, humidifiers, and heaters. And we’ve flattened neighborhoods, carved tunnels, and created elaborate fueling networks for the automobile. By our own definition our homes, neighborhoods, and cities are robots we’ve never bothered to name.

    Which extends the original definition into meaninglessness, if we were to rely solely on a technical definition. By pushing the assumed boundary to the extreme, we’ve uncovered the real divide that angers roboticists worldwide: conflating the technical work of the word “robot” with the cultural work it’s doing. Technically, a dishwasher is indeed a robot. But culturally, it’s an appliance.

    A technical definition provides a model of what a thing could be classified as for the purposes of figuring out which rules ought to apply to it. Straightforward, simple, and entirely useless for the cultural work that a word needs to do. As human beings, we use the word robot colloquially as an othering term for new technologies. These othered technologies are new, strange, scary, and threatening, here to take your place in society and knock you down a few rungs economically and socially. Once the othered technologies become familiar, and we see that they’re net benefits and not the destroyers that we thought they were[1], they earn names of their own.

    So take a look at what we still call robots: they’re strange new worlds of technology, early in their journeys within our societies. Roombas are mid-journey: familiar and approachable enough that they’ve begun to be anthropomorphized with names, attributed personalities, and even microfic[2]. Dishwashers have completed their journey to ubiquity as appliances.

    All great robots eventually take their place as a named appliance. Dishwashers just got there first.

    [1] – Of course, excluding when such “others” are used as socioeconomic weapons, but that’s a discussion for another time.
    [2] – See the ever-growing body of work on “Captain Stabby”.

  • Hit Dogs Bark

    So a couple weeks ago the FCC dropped a National Security Determination and I had to go home, make popcorn, open the last of the random rosé I had (truly an event for a gin drinker), and watch the robotics zeitgeist have a complete meltdown. Oh, dear reader, I have not hate-scrolled the socials like this since Kendrick nuked Drake.

    For those of you who have thus far escaped the drama, the Trump administration dropped this doc on 27-Jul that puts mobile robots and power inverters on the FCC Covered List, putting them into the same regulatory category as Huawei and ZTE. The scoping turned out pretty sensible, excluding anchored robotic fixtures, medical devices, and drones (which are covered separately by the FAA). There’s a Conditional Approvals path for foreign manufacturers to demonstrate provenance and gain exceptions.

    And then the entire robotics takes-having industry lost their fucking minds. They’re killing American innovation! They’re trying to screw over China! HIDE YOUR CHILDREN, THEY’RE BANNING YOUR ROOMBAS

    Granted, the committee that drafted this document did not cover themselves in editorial glory; the lack of care in its writing certainly doesn’t lend itself to credibility, but the core is sound. The only good analysis has been from my favorite technology hot take podcast ChinaTalk, where they go in depth on the intent and how the FCC mechanisms work. Particularly noteworthy is the insight that the FCC Covered List actually makes a LOT of sense for regulating emerging technologies with shorter expected operational lives.

    So let’s get into why everyone really freaked out: the NSD called out actual sins of the industry. Here’s a few:

    • February 2026: a security researcher found a vulnerability that gave him remote access to 7,000 home robots simultaneously — live camera feeds, microphone audio, and detailed maps of consumers’ homes.
    • September 2025: an exploit allowed a remote actor to take over entire fleets of Unitree robots.
    • April 2025: security researchers found a potentially pre-installed backdoor in Chinese robot quadrupeds providing access to the camera feed and full remote control of the device.

    Documented incidents in named publications with verified sources demonstrate a trend of very real industry behavioral problems. No one in the administration made this shit up to punish China or go after Unitree or some other nonsense. Pairing the robotics and power inverters actions in the same NSD makes it clear that this isn’t about geopolitical maneuvering; it’s genuinely going after long-standing InfoSec/InfraSec concerns. The Big Bad Government read the news, extrapolated appropriately, and did their freakin’ jobs.

    Chat, regulations are written in blood. The pattern’s the same every time: an industry wings it on their own for a while, generally with the best of intentions, sometimes making mistakes but building up a body of common best practice knowledge. A subset of profiteers then cuts corners in the name of time-to-market or competitive pricing or straight-up sloth, and people start getting hurt. Eventually, enough people get harmed, maimed, or killed that an uproar finally forces action. This is how we got damn near every executive agency, from the FDA to the EPA to OSHA to CISA, because a single mere mortal cannot possibly perform enough due diligence on everything that we touch and consume to correctly judge the risk levels. And even when they could, incentives like “needing to pay rent” can force people into situations where they have no real choice.

    Which tells you exactly how to read the reactions. Anyone who read and understood the NSD but howled anyway probably is not thinking about safety and security the way someone who makes equipment that can harm, maim, or kill ought to. Regulations and standards accelerate organizations doing the right work because they provide a roadmap on best practices and a ready-made guide to what “good” actually looks like. These organizations welcome scrutiny because they know it’ll only make them better.

    The screamers fear scrutiny because they either don’t know about or haven’t bothered to identify the gaps between what they’ve produced and what good is. The ignorance is fixable, and regulatory requirements before you can go to market help do so significantly. The ones who don’t bother are the scariest, happiest when they can push any and every cost onto everyone else around them no matter the consequences.

    Those latter ones can be prevalent in any industry with messianic hype complexes[1]. We are absolutely infested with amoral longtermists who aren’t afraid to shed our blood today for their benefit tomorrow. Regulation like the FCC Covered List stops these fuckers from making it everyone else’s problem.

    So a couple things for the to-do list from here:

    1. Go back and read the NSD again because it’s telling you what good looks like: supply chain provenance, software integrity, decision-making safeguards, and responsible data handling. Your Product people should just write goals and requirements against those right now. It’s a gift.
    2. We know how to demonstrate what good looks like, so let’s get it into real regulatory policy outside of an NSD memo intended as a stopgap. Our industry needs to stop waiting for spankings and lawsuits and build our own proposal to put into US Code. Modeling how Title 21 sets cGxP for pharmaceutical manufacturers with appropriate tiering for the levels of risk, and governed by CISA which has the analytical capabilities to assess it, provides clear public guidance so that we all know what the bar is and how to hit it.

    Stop barking and get to work.

    [1] – see Move Fast and Break Accountability for more on how we keep ending up here.

  • Move Fast and Break Accountability

    The further you are from the outcomes of a decision, the easier it is to make a bad call. It’s a natural consequence of large systems where the decision-maker gets delinked from the person who has to absorb the results. And because it’s a systemic problem, that means we can design it out.

    Two things that ruin my day: arbitrary bureaucratic hurdles and being told to “go fast and break things”. Turns out, they’re both just the extreme ends of the same risk spectrum: too risk-averse and you don’t get anything done, too risk-happy and you hurt people.

    The problem is, localizing yourself on that risk spectrum requires the right incentives. It’s how you end up with both symptoms exhibited by the same team: be bold, but I don’t want to get yelled at if something goes wrong.

    Let’s start with some definitions.

    First off, bureaucracy is not a dirty word. It’s the overhead that makes a large people system with many actors produce consistent results at scale without harmful results. It’s a force multiplier.

    When you go too lean on bureaucracy you get to watch engineers trying to reinvent things like program management from first principles. Good bureaucracy puts structure around decision-making and ensures that the right people are having the right conversations at the right times. Bad bureaucracy induces valueless make-work. People who act like children often point fingers at good bureaucracy when it has prevented them from enacting a stupid idea. This sort of whining accounts for about 60% of the content on AM radio.

    Now, “moving fast and breaking things” isn’t bad either. You’re giving yourself permission to do something differently that may require other changes in the environment going forward. But remember that “move fast, break things” explicitly calls out the price of doing so. What things are you breaking? Can they be unbroken? What if the things you might break are people? Furthermore, what if you’ve abstracted away the things you might break far enough away that you can’t even see them when you’re moving fast?

    And how do you find the sweet spot?

    My risk management peeps already know what’s coming, but rather than asking you to FMEA your life here’s some simpler questions instead to start figuring out if you’ve got it right:

    1. Can I even describe what “going wrong” might look like? Recovering juvenile delinquents and budding horror authors, this is your space to play in. Get imaginative.
    2. Can I quantify that in terms of likelihood? Quantification turns uncertainty into risk and means we can grapple with it. One in a million hits a lot differently when you have five million transactions an hour.
    3. Can I hit the undo button? Basically, how reversible is this? If it’s a software rollback with no data destruction, easy peasy. If you’re an autonomous vehicle trying to decide if that’s a toddler or a shadow, not so much.
    4. Who gets hurt, and did they get a vote? The lady in the warehouse standing next to a robot wasn’t in the sprint planning meetings. The guy getting his medical claim denied by his insurance didn’t have input on the model weights. Yet people very far away from both of them accepted risk on their behalf.
    5. Does whomever made the call have to face the consequences of the outcome? Accountability rolls upward to human beings, not downward to machines. If you put in a decision-making system, you own the decision. And the further away you are from those consequences, the more careful you must be.
    6. How do you know the right decision was made? Detectability is key here for any decision-maker capable of learning. You can’t change how you make choices if you don’t get any feedback.


    
Setting aside the psychos, decent well-meaning people get the assessment of the above wrong all the time. Four main factors can do it to you: decision fatigue, blame avoidance, self-preservation, and short-termism.


    Decision fatigue. Every decision requires cognitive energy. It’s why you hear of so many leaders emphasizing their monolithic wardrobes: even choosing your socks costs a decision. Furthermore, context switching costs even more energy. When you’re out of juice, you default to whatever’s safest for you, including anything that conserves that cognitive energy.

    Blame Avoidance. No one wants to be yelled at. Delay costs nothing, whereas a perceived bad decision costs uncomfortable meetings. You’d rather optimize for whatever your boss thinks looks good versus what you might know is actually good.

    Empire Building. Everyone wants to keep their jobs, no matter how phoney-baloney they may be, because everything else is predicated on having a job so that you can live. No one got an incentive for making their own existence go away.

    Short-Termism. If you’re not around when the consequences land you have no incentive to do anything but discount the risk to zero. When you receive positive incentives for an excellent quarterly return, but the consequences hit in year 3 long after you’ve moved on, why not claim the win and run? Longitudinal studies rarely follow decision-makers around.

    Now, are any of these novel concepts? Hardly. There’s a reason these all have such familiar names. We’ve been at war with these in human organizations since we started being human. Decision-making systems are decision-making systems regardless of the actor.

    This is an allegory about agentic development.

  • Amazon’s Writing Culture Is Eating Itself

    Faster document production is putting us into a decision quality doom spiral. Let’s talk about how two trends are killing us cognitively and what to do about it.

    The outside world often makes much of Amazon’s writing culture. You can find plenty of writing on it from others, but the basics are: all proposals, reports, and reviews are presented in the form of a written narrative document, no longer than six pages (though you may add as many appendices as you like). We read the document in person for the first portion of a meeting, then discuss afterwards.

    While we gain the advantages of building a collective, durable memory and the ability to share ideas outside of an ephemeral meeting, most importantly this approach promotes critical thinking on both the input and output sides of a proposition.

    On the input side, the author cannot “baffle them with bullshit” in a rapid-fire slide presentation. They must think through the problem, consider the options, and then construct the story in a concise and coherent manner that stands up to scrutiny outside of fast talk and shiny slide transitions. Paper writing helps authors find the holes and inconsistencies in their own concepts, making for more thoughtful and thorough products.

    On the output side, the readers have time to digest the “why’s” and “what’s” of an argument. They can go back and forth between sections and see the connections, gaining the nuance provided by the richer context. This not only allows the readers to form better critical questions, it engages their minds fully in the story and situation to allow them to make better decisions.

    But there are two trends I’ve seen appear in the past few years that subvert the intent of the document-based culture and, I suspect, are contributing to worse outcomes: AI-assisted writing and Page Zero metrics.

    How AI-assisted writing in its most extreme form subverts the critical thinking of the author should be obvious. Feeding some bullet point thoughts into an LLM and instructing it to turn it into a white paper just takes a slide deck and makes it worse. You’re meeting the form and losing the function.

    Generally speaking, when I suspect that a white paper has been primarily authored by an LLM[1], my brain turns off. If the white paper presenter doesn’t respect me enough to spend the time thinking through their argument, I’m not going to respect the slop they’ve put in front of me. It’s the only time I’ll use an LLM to summarize a white paper for me.

    Now, LLMs can still be useful in drafting white papers. I’ve found them to be effective in “linting” a white paper for readability and voice. They’re most useful when used as a mirror, to reflect on your arguments and whether they’re coming across properly. But that’s after the primary writing is complete.

    The other trend, which is likely more Amazon-specific but far more dangerous, is Page Zero metrics. They’re just what they sound like: a “cover sheet” of a big pile of numbers before you get to any of the narrative behind them. An ocean of green up arrows and red down arrows short circuits any nuance and obviates the value of the following narrative.

    Leaders like Page Zero metrics because they’re easy to digest. If you see a red arrow, you can nitpick it and feel productive. Just make all the red arrows green and you’ve done your job, right?

    I recently spent 21 minutes of a 60 minute meeting listening to leaders bicker about whether the source of a certain metric was sufficiently accurate, when the metric is actually a proxy for whether the customer could achieve their goal. These leaders completely missed the real conversation they should have been having that actually was covered later in the narrative, but they hyper focused up front on that little red arrow and that was that. The much more nuanced and scarier conversation about end of life components that the team really needed executive decisions on was completely lost.

    Page Zero metrics have an easy solution: go back to basics. Quantitative measurements do still matter, and they need to make an appearance. But Page Zero metrics aren’t a problem of data, they’re purely a problem of information design and crafting the conversation you need to have.

    So put them in an appendix and drive the conversation as going section-by-section from front to back. Feature the conversations you want to have at the front of the document. If your leaders have too much fatigue from switching from meeting to meeting all day and can’t focus, go the psychological route and find the ways to make the document easier to digest through structural, grammatical, or even scheduling approaches.

    If you can, kill some trees and have your readers review an actual physical white paper in person, using a pen to make notes in the margins. The physical interaction further reinforces engagement with the document, focuses people away from “multi-tasking”, and breaks your readers out of screen hypnosis.

    Using written documents to interact with ideas provides the framework for the critical thought that LLMs can’t reproduce. We can use tools and mechanisms to improve the quality of our outcomes, or we can cut corners and turn into a cargo culture that preserves the ritual of the white paper while losing the thinking it was intended to produce. Don’t shortchange your readers, and don’t shortchange yourself.

    [1] – To be clear, certain markers on their own don’t indicate LLM authorship to me per se. For example, African writers tend to get falsely called out by American readers for phrases such as “delve into”, which is really just an artifact of the British English they learned in school. And I love my em-dashes. You’ll never take my grammar fascism from me.

  • We need to talk about humanoid robots.

    Humanoid robot shipments are up in 2026, there’s a multiplicity of vendors out there, and high production demo videos abound. Yet, no one’s asking the most important question: “So like, what does it actually do?”


    They definitely look cool, and I’m enjoying watching the capabilities develop and some of the technology trajectories that they subsidize. But while some companies may be claiming early commercial deployments, so did Juicero. It’s going to take thinking through two dimensions for them to be able to go pro.

    Ex-Kivans will know these dimensions by heart. The first is, of course, Mick Mountz’s axiom that “it’s not about the robot.” The second is Pete Wurman’s exhortation to “do one thing, and do it very very well.”

    
As human beings ourselves, we’re naturally in love with the human form. It’s a major factor in why humanoid robotics is literally all about the robot. Look at them walk, run, dance, make faces! See how closely our manipulator mimics a human hand! Great technology demonstrations, all of them, and on individual metrics they out-perform human beings.

    But those individual, isolated metrics don’t tell us anything about how a robot works as part of its larger environment. The value of automation, be it a dishwasher or a humanoid robot, lies in whether it has been correctly matched to its task and integrated well into the megasystem embedding it. It’s not about the robot, it’s about using whatever correct tools are available to best solve a problem repeatably, reliably, and at scale.

    But like Pete said, it’s to best solve a problem, not solve all the problems. Human beings are generalists, which means we’re kind of okay at all sorts of different things. To be viable enough to be worthy of sacrificing human creativity and intuition in a process, a robot (like an insect) is a specialist. When you ask a human being to do a robot’s job, our brains go to mulch and our bodies get slowly ground away. When you ask a robot to do a human being’s job, you’ll have an endless supply of unsolved or mis-solved edge cases to clean up.

    The human form of humanoid robots lets us mislead ourselves that they can replace the holistic human being in a system without thinking more deeply about whether the process itself should be redesigned.

    “But the world is built for humans!” And yet time after time, when we find automation that’s sufficiently useful we gladly modify our environments to incorporate it until the modification itself becomes so familiar as to be invisible. Elevators, railroads, and automobiles required far more drastic changes to how we design our worlds. If you’re afraid that your robot isn’t worth modifying its operating environment for, you’re probably not looking at the problem or your assumptions correctly and are trying to shove a robot into someplace it shouldn’t be.

    So… are humanoid robots going to be the next smartphone or the next 3D TV? One found its killer use case and became infrastructure; the other solved a problem no one had and died a historical footnote. If you’re working in this area, your Prime Directive must be forgetting about the cool factor and finding the one place that humanoid robots can do a thing— regardless of how unsexy it is— better and more reliably than anybody else.


  • Tina’s Maxims

    Tina’s Maxims

    This post was adapted and updated from my original posted on Medium.com in 2022.

    I’d like to share the maxims I have developed over the years to guide my professional approach. (I’m not sure my colleagues would dare label any of my approaches as “professional”, but I digress.) These probably each deserve their own essay at some point but I figure you could at least use a little taste, as some sort of perverse corporate treat.

    I use a simplified bullet journal at work. It’s my external brain for notes, to-do’s, and information I want to keep at my fingertips. Among those are a set of maxims I’ve developed. While some pages in my notebook have been cut and reinserted into successive books over the years, the Maxims page is one I always rewrite by hand. It’s a great opportunity for wordsmithing and measuring whether the maxim is still valuable.

    Maxims describe fundamental principles of behavior. They’re different from tenets, which, when used correctly, provide a fulcrum for decision-making. Both are difficult to write, and both must be continuously revisited, revised, and maintained.

    There’s no ordinality here beyond order of initial development. A couple are similar in nature. All are irreverent and not intended for shiny, slick slide decks about How To Product.

    Tina’s Maxims (as of mid-2026)

    1. Averages of averages are Satan’s butthole.
    2. Achievable != sustainable.
      • Possible != feasible.
    3. All band-aids immediately operationalize.
    4. Does it scale? Up and down?
    5. People will misuse your products in ways you cannot possibly imagine.
    6. Design like your users are drunk.
    7. Document like you’ll get hit by a bus.
    8. Slow is smooth. Smooth is fast.
    9. Killing a bad feature is at least as valuable as launching a good one.
    10. WRITE. SHIT. DOWN.
    11. Don’t leave money on the table.
    12. All models are wrong. Some are useful.
    13. In optimization models, only Sith deal in absolutes.
    14. Fear is the mindkiller. Anxiety is its kissing cousin.
    15. Perfect is the enemy of good.
    16. ALWAYS FIGHT BULLIES.
    17. “Not looking stupid” is a feature.
    18. Assholes are rarely assholes in only one dimension.
    19. You cannot optimize a system you do not understand.
    20. Do not confuse motion for progress.
    21. With enough money, anything is retrofittable.

  • Hello, world.

    A place for the Robot Girl Gang to share things that The World Needs To Know.