A simple explanation for why “estimations” and timeline-based “roadmaps” don’t work, is that product development isn’t a scheduling problem but an explore/exploit problem. Time isn’t the constraint, but knowledge. Knowledge that isn’t currently “priced in” to the market.

I don’t fully buy the reason that pinned tabs, vertical sidebar, and spaces are “too complicated” for normies We live in a world where even mainstream browsers (Safari & Edge) have sidebar tabs, collections, synced tabs, and pinned tabs Heck, Chrome has had pinned tabs since a decade ago I think the reason Arc feels complicated, honestly, is that they’ve focused a bit too much on that circa 2010s hipster aesthetic and gaudy visual design that gets in the way of usability When normies don’t understand your product, you run a usability test and fix those issues. You make your visual design less avant garde and more utilitarian. You tone back your marketing from being aspirational to being pragmatic and utilitarian. They’ve done none of the above, and in fact are doing the opposite! Still think it’s crazy that they’re abandoning the product instead of doing those things. Imagine if Apple decided to ditch iOS and iPhone completely instead of redesigning iOS with flat design and adding features that are inspired by Android. When they could’ve simply cleaned up the design to be cleaner and less childish.

The difference between Rabbit, Humane, et al and Apple is Apple cares about human interface & ergonomics. I don’t see any of the longstanding HCI challenges addressed by any of these newfangled companies. They just blatantly ignore HCI best practices and assume they know best.

Did Apple just solve authentication? #SignInWithApple #WWDC2019

Never got why Vercel throws so much weight behind a “server-first” approach when local-first with background syncing and conflict resolution is the winning approach. It’s as if they’re financially-incentivized to care about a less optimal approach, or am I missing something?

Replacing graphical UIs with natural language queries isn’t better by any means. Because it means the user now has to recall exact tasks and workflows, instead of having a fuzzy goal in mind, then pick from a menu of options which they recognize, which is cognitively easier.

Math Notes is the most Bret Victor thing that Apple’s shipped, in recent memory. Direct manipulation 💯 Handwriting ↔ Charts ↔ Variables coexist in a single coherent medium that doesn’t just mimic paper, it’s computationally-rich.

There seems to be 3 primary trends in this new breed of productivity tools: 💎Luxury: @Superhuman, @linear_app. Your time is precious! ⚖️Egalitarian: @NotionHQ, @figmadesign, @MiroHQ. Work is better, together. 🔪Brutalist: @RoamResearch, @coda_hq. No pain, no gain. (Thread)

Has anybody contemplated taking or have taken a mini-retirement? Not a long holiday but a hard career break to figure out next steps. 2020 has left me rather burned out, and I know of few people that have taken the leap.

Why do models for organizing work in tech companies fall into either the “mission control” (auftragstaktik) or “command and control” categories? Those are borrowed from the military. Why not models borrowed from the arts? Like a TV writing room, or improv, or jazz.

Get Pock. pock.dev/ Let’s you show your dock, the time and date, your battery level, and the title/artist of the current song. Best part is, it’s stable and doesn’t change from app to app. Can’t live without it now!

Great quote: “The great horizontal killer applications are actually just fancy data structures”. I would add: it’s not just data structures, but information representations that expose those data structures in an intuitive way. Intuitive how? In that the graphical representation and interactive controls provide a transparent interface to the underlying data structure, where: - its graphical representation easily communicates that data structure’s current state, and - its interactive controls allow for a wide range of operations for manipulating the data structure’s state This idea of Excel not being widely used as a calculation tool, but a way to lay out lists and tables, seems to suggest that most people see software first as a medium, then a tool, secondarily. A medium lets you shape information into specific representations. Different representations help you externalize information in a different ways that facilitate certain styles of thinking: A Kanban board lets you visualize tasks in a left to right fashion that represents the status of a task, so you can visually grok progress. The columns let you see how “full” your workload is, how congested your pipelines are. A diagram lets you see relationships between parts of a system, or between concepts, more easily. Without which, you would struggle to think non-linearly. A table lets you think of your data in two dimensions. Yes, some of that data could be quantitative in nature, and so this medium lends itself well to calculation, aggregation, and charting. But before you can manipulate data, you first need to be able to think about that data. That’s why spreadsheets still feel more natural to use than SQL reporting tools: those tools conspicuously lack a medium for which data can be represented as useful information first and foremost.

Something I’ve noticed. Once you become even halfway as knowledgeable as “the most knowledgeable”, your perception of who’s the most knowledgeable in the field changes. You realize the ones who appear to be knowledgeable to beginners are frauds who repeat buzzwords for a living.

These companies are trying to replace GUIs that have been designed with humans & our cognitive limitations in mind, with newfangled UI that hasn’t. “Recall over recognition” is a bad design principle And who wants to deal with the Principal Agent Problem when using software?

Replacing apps with “agents” (even if they work perfectly & don’t hallucinate) means the user has to deal with results that drift from their intent. This is okay for interpretive & creative tasks like music, video, or image creation. This isn’t okay for productivity.

Took a long time to realize I should be reading Taleb, Hayek, Judea Pearl, Donella Meadows, Ray Dalio, etc. Instead of what is usually recommended reading for product managers like Toyota Way, or Marty Cagan.

It’s honestly rather surprising that software designers aren’t spec’ing out designs that consider the entirety of the system they’re designing. Mental models for planning how software systems should work, like MVC (model-view-controller) and React’s uni-directional data flow where UI = f(state) have been around for a long time! Yet designers often only work on the view layer. Why not design the model layer too? (information architecture & data modeling) Why not design the controller layer too? (Spec’ing API endpoints and logic) The split between software designers who work only on the view layer, and software architects, who work only on the model/controller layer (often not communicating with each other except through a frontend engineer or product manager) might be why so much software breaks so easily. When the design of your software is siloed between different practitioners who cannot speak the language of the other, it’s no surprise that the result is an incoherent mess.

Product development is closer to stock trading & epidemiology than manufacturing or traditional pure software dev. 🤷‍♂️ What confused me when first starting out in product management is learning techniques rooted in manufacturing or software dev, then struggling w them in practice

Folks who work on cross-functional product teams, how do you run retrospectives? What format are you using, and how did it get this way? Looking for inspiration beyond the “what went well/what didn’t go well/what to improve” variety.

Designers say this all the time: “Open-ended text boxes for prompting are bad design because they don’t communicate any affordances to the user” But without realizing that the open-ended text box is the affordance! The “blank canvas” feeling you get with that empty input box is a feature, not a bug. Generative AI tools are used for learning, creativity, and productivity. Is there a tool that supports these types of tasks that doesn’t start with an empty canvas or document? Find me one. I’ll wait. Notion/Google Docs: empty document Trello/JIRA: empty board Figma/Photoshop: empty canvas VSCode/XCode: empty file Jupyter Notebook: empty notebook Ableton/Logic: empty timeline Premiere Pro/Final Cut Pro: empty timeline I could go on and on and on. Somehow ChatGPT starting with an empty chat box is problematic? Surely that’s wrong.

I love it when companies publicly share their “ways of working”, especially if it’s nuanced and comes with disclaimers of how that way might be challenging. @TransferWise does a really good job. Does anyone know of any other companies that do this? tech.transferwise.com/produc…

Quite often, software products that people adopt are just the product they switched away from, but with better UX. They don’t need to be “special”! @cursor_ai = VSCode with better UX @arcinternet = Chrome with better UX @figma = Sketch with better UX @typescript = Javascript with better UX @nextjs = create-react-app with better UX @supabase = PostgreSQL with better UX It’s a no-brainer deal for the customer. It’s the same product, but better! You have everything to gain and nothing to lose. This sounds like it shouldn’t work, because the product that people are ditching should recognize this is happening and just copy the UX improvements which would make the new incrementally-better software wither away in obscurity. But for whatever reason, incumbents don’t bother reacting to their better clones, until it’s too late and almost everyone’s already switched. The one notable exception is products with lock-in, like social networks and platforms with closed ecosystems. New products challenging products with network effects or lock-in can’t win adoption on better UX alone, but even so, challengers can succeed! @NotionHQ is Atlassian’s suite but with better UX, and they’re winning. @instagram stole enough users from Facebook that they were acquired.

Perhaps the worst thing about dashboards is how you can’t explore / play around with the data and ask questions. There’s no way to “inspect source” to see where data comes from or what it means (often more poorly documented than a DB). They’re dynamic, but going only one way.

What we need is better ways to steer foundation models to more correct, precise, and coherent outputs. What we got is worse ways to use software for productivity-oriented tasks.

Nobody rebranded anything. The term emerged from the library that was used to create decentralized Ethereum dApps, and became used as a generic term to describe those dApps, and later became broad enough to describe similar apps on other blockchains with similar architectures.

Finite state machines as a first class citizen in design tools would be nice. So much of state management in UI dev is UX-related (loading, timeout, empty, error, placeholder, pressed, expanded states, etc), yet we leave it up to software engineers to decide.

“A network of questions we want to answer” – a more nuanced and transparent strategy, and a more accurate roadmap than what most people call roadmaps. A roadmap is a diagram which shows a network of roads (and not necessarily even the path through them).

Yeah I don't like the avant garde UI. This isn't the 2010s anymore. Standards for UI have kind of standardized on a few patterns, and for good reason. If their new product is as funky as Arc, they won't be getting anywhere close to the adoption they expect to be getting.

Facebook rebranding to “Meta” and Square rebranding to “Block” is so gauche, like if IBM rebranded to “GUI” and Oracle rebranded to “Cloud”. Companies trying to prematurely hitch their wagon to a future tech without proof of work. “Concept-squatting” might describe it.

I’m a designer. This isn’t a critique. It’s me thinking aloud. By articulating my opinion on design choices in the wild, and conversing with others about it, I’m hoping to develop clarity of my own point of view.

Love this way of thinking about the Web, not as not one but two axes: online vs offline, static vs dynamic Helps me realize why I’m frustrated with the landscape of web dev frameworks. I want tools that focus on local first. But what does that mean?

If you render your UI on the server, you’re putting your server in the critical path of UI rendering. For productivity apps with frequent interactions, it’s just nuts to do that. I tried using a productivity app built with the Next.js app router, and it was pure torture.

This app @rainbowdotme is killing me. I haven’t had so much fun with a crypto app. 🌈

In this two-parter blog post, I share how the mobile product team at ⁦@tenxwallet⁩ rewrote our apps in ⚛️ #ReactNative, ⁦@Expo⁩, ⁦@typescriptlang⁩, and ⁦@apollographql⁩. It was a blast seeing our designers ship code to prod! 🚀 blog.tenx.tech/react-native-…

Feels like this style of software could work only if local-first and/or multi-tenant databases properly take off. LLMs alone won’t unlock malleable software. Yes, AI can write code. But SaaS software these days are built such that data is server-authoritative, and they implement tenant isolation at the application level. How would an LLM even begin to modify software to each user’s liking, when the data and business logic between tenants is commingled between all tenants? How would an LLM: - Migrate the database to a new schema? - Create new APIs that talk to that database? - Make customizations that are additive to what the app developer already built (like a non-destructive patch?), so that it doesn’t conflict with what the app developers will later add? - Extend or modify authorization logic that already exists in the base system? (Enterprises care about this!) Sure you can design around these, but it tends to end up looking like Figma plugins or JIRA apps, where the app itself is untouched, and there are narrow integration points in either the UI or API layer you can hook into. Actually modifying the behavior of the core SaaS app sounds like science fiction to me. Until there is a reliable off-the-shelf sync engine and non-proprietary way to implement multi-tenant databases, it’s hard to imagine building an app that lets users customize beyond the UI layer. Sync engines let the database be automatically replicated to the client-side. So customizations that don’t require a change to the database schema can be applied completely client-side, eliminating a huge class of potential issues. Multi-tenant databases let customizations be applied on the database schema, if necessary, but those changes can be isolated to a single tenant. Cool that YC has put out a request for startup for this idea though. We might start to see some progress here!

Lots of ink has been spilled over this, and most books written about this make the mistake of comparing software development to manufacturing. Software development is R&D, not manufacturing! Just ask your accountant. Of headcount-related expenses in R&D, software engineering is probably the highest of them all. Makes perfect sense that there would be pressure on software engineers to estimate their work in terms of time. But of course, any sane person would realize that “how long something takes” cannot be estimated in any sort of R&D activity, software or otherwise. Unless, of course, there’s no R&D involved at all. Routine maintenance work and cosmetic changes can be predicted well ahead of time, as well as UI engineering, DevOps, infrastructure engineering, and data engineering which are more manufacturing-shaped than new product development. A software development culture that values time estimates will inevitably trend towards manufacturing-like activities. In other words, you’ll see “scaling” work and “redesigns” but never any sort of innovation. Almost anyone who has worked in software knows that companies, as they grow larger, tend to innovate less and less. This is probably the reason! A focus on manufacturing-shaped activities, rather than R&D-shaped activities. Real stupid, in my view. Your tax accountant and the tax authorities still consider software engineering R&D, so you’ve pretty much played yourself. Not to mention that competitors — startups, indie hackers, and open source projects — who still do software development as R&D, will eat your lunch.

None of them, yeah. Human achievement isn’t zero sum like that.

Basic things are still too difficult and not baked into the platform (i.e., web browsers): Auth, real-time collab, local-first, data reactivity, tooltips, modals, text editing, payments

When I play a multiplayer video game I expect the world (map, models, textures, etc) to be loaded locally from disk. And the state of the players to be synced in the background over the wire. It would be super weird if the map was streamed together with player state!

I’ve lived elsewhere, have known others who have, and many others who’ve moved to SG too. While I’m aware there’s many things Singapore does “better”, there’s also a deep realization of the social costs it took to get there. Esp when you have family members who are less well off

Have been working with startups for most of my career. One of the things I’ve been thinking about lately is the concept of Founder-Domain-Intervention fit. (Domain instead of “problem” or “market”, and Intervention instead of “solution” or “product”) Thread 🧵

2015: On-demand grocery delivery 2020: Standing in line for a “virtual queue” into an online grocery store

If the demo below works as promised (graphics ↔ code isomorphism), then this is a watershed moment for screen design. Where you can design with tight feedback loops against a real artifact. Something musicians, video editors, illustrators, print designers have had for decades.

Product research tip: You can deduce a lot from existing models. Whether it’s heuristics used by experts in the field or formal theories. It’s rarely necessary to go straight for empirical research (i.e. interviews). Unless you feel the existing knowledge is lacking.

Why do LLMs make schemas, fields, etc obsolete? LLMs’ big weakness is that they tend to hallucinate facts. They don’t really “know” the underlying structure or semantic meaning of text, only what’s likely to come next in a sequence.

I think it’s because thinking of domains and how they’re split or named is hard, and needs upfront thinking It’s much easier to create a utils, hooks, or types folder and throw stuff in there

It is quite sad. The consensus of what “research” means to many designers and even PdMs is to rely on empiricism alone (i.e. interviews, surveys, usability tests). Before you spend on new empirical studies, you ought to do some desk research on prior literature lol.

Many don’t even know the differences between research methods, their cost/benefit, and when to use which. Observational empirical studies like interviews are the most expensive and the least actionable. But they are likely to draw out surprising information (unknown unknowns).

What seems under-explored in the realm of social networks is the design space that lies somewhere between group chat (Discord, Slack) and broadcast-posting (Twitter, Farcaster, Mastodon), especially along the lines of in-group reputation.

Local first tech is the most exciting And multi-tenant databases

Seems like they’re just running some kind of headless browser in the background after you auth into your services, to take actions on your behalf. I think they can offer an app with a few Siri Shortcuts exposed, I don’t see how Apple would be a roadblock here.

I don’t know much about Art, but I know one thing: Art derives its value from scarcity Right now that scarcity comes from an artist’s intent, the artwork’s place in history and culture, and the story of how it was made, where it was exhibited, and what critics and scholars thought about it Its “provenance”, if you will When someone makes a bold claim like “X can’t be Art”, I immediately think of the question, “Why can’t X be imbued with an artist’s intent, be historically or culturally significant, and be exhibited and collected as an artifact?” It stands to reason that any output of an AI model can be imbued with an artists intent, be historically or culturally significant as a result, and be exhibited and collected as an artifact. So it’s unreasonable to argue that outputs of AI models can’t be Art. If a urinal can be Art, and a banana on the wall can be Art, and stencil graffiti on a wall in Bristol can be Art, why not any arbitrary bunch of pixels? What Ted Chiang is arguing, which is that “AI lacks intent”, well, that’s true for every medium that was used to create Art in human history. Marble lacks intent. Photographs lack intent. Oil, canvas, acrylic all lack intent. The intent belongs to the artist who works with the medium. This isn’t new or surprising! There’s also the argument that prompting isn’t great at representing an artist’s intent, because right now image generation models aren’t very good at adhering to prompts. It’s mechanically producing an output based on statistical probability. But that’s not an argument against AI outputs being Art. An artist takes an object that was made for some other purpose, or made by a mechanical process, and presents it as Art. Is it Art? The answer is “yes”, and that’s been the consensus for more than a century now. Few artists, collectors, curators, scholars, critics, or restorers would argue that an arbitrary “found object” or a mechanically-produced “readymade” can’t be Art. tate.org.uk/art/art-terms/f/… tate.org.uk/art/art-terms/r/… So why is an outsider to the art world like Ted Chiang arguing otherwise?

Zen and the Art of Motorcycle Maintenance Thinking in Systems Algorithms to Live By Asimov’s Foundation & Robot series

Feels like people keep circling around the same ideas that revolve around avoiding meta-work (management of tasks, references, and ideas) “Let the computer remember it and structure it for you” But maybe the value of that work lies in its doing

Tried this browser yesterday. I know such UI has been prototyped before in the past, but never really made into a real browser. Pascal is turning it into a real one! It’s really nice, actually. It’s like a Roam, Workflowy, or Tana style outliner, but nodes are browser tabs.

I have this yearning too. Feel letdown by tools like Notion and Miro, and graph-based PKMs. For a good few years it seemed like they might make progress, but then at some point made a left turn into enterprise tools or “creator” tools territory.

What we’re facing right now is like the plot to The Expanse or Attack on Titan but without the protagonists trying to fight for humanity.

Beyond the first load, offloading rendering to the server feels wrong, UX-wise. The justification that “the server needs to do work anyway, why not also render UI?” feels iffy because UI isn’t data. UI is the “place” where data lives. That place should be persistent.

Where did that talent go? 3D modeling, probably. Look at how life-like a game running on consumer hardware from 2018 looks. It’s not just sculpting human forms anymore. Humans have mastered modeling in 3D space anything you could possibly think of, and it’s commoditized.

And the distance you thought was the midway point is now probably a single digit % the way there, because new horizons appear to you that fake experts obscure. So it pays to get there ASAP. If you can’t call BS on the most popular experts of a field, you haven’t understood it.

Everything is a token (LLMs)

Not sure what this style is called. Digital Analogue? Retrodigitalism? Arturia, Ableton, Teenage Engineering kind of adopt that style in real-world products. Inspired by retro computers. White & black with restrained color, black screens w/ single-width lines. Knobs and dials.

In the other universe, they call us the “anti-universe” too.

I can live like a globalist in so many cities around the world – New York, London, San Francisco – but there’s nowhere like home. Except, my home no longer feels like home. It feels like a poor imitation of a Western city. So yeah.

“Crypto” as a term has been around for a long time going back to WWII when it was extremely prominent. Most recently we converged on cryptocurrency as the latest novel cryptosystems, but nobody “branded” that either, it emerged from decades of use as a short form of cryptography

web3 apps are a subset of crypto. When we talk about web3 we refer to dApps that have a specific architecture, the protocols that support it, and the community that builds and uses those apps and governs those protocols.

Some examples of graphs used as a strategy / planning tool: 1. Wardley Maps by @swardley 2. Opportunity-Solution Tree by @ttorres 3. Impact mapping

Having a persistent model of a network of beliefs > isolated hypotheses. Network of beliefs helps you do this: 1. When A is true, B will be true. 2. When B is true, C is true 75% of the time. 3. A is true, so C is most likely true.

All this talk of “digital property rights” is about avatars, game items, JPEGs. I can’t even bring my podcasts, audiobooks, music library, ebooks, bookmarks, notes, highlights, photos, calendar events, messages, contacts, from one app to another without “exporting” or “syncing”.

Glad I moved my primary knowledgebase to @tana_inc They’re the only team focused on power-user features with a design bent. I like Notion, but they’re no longer focused on power-users. Roam & Obsidian have power-user features, but aren’t design-focused.

🤔 One thing I’ve always wondered… why aren’t graphs or maps, as information structures, more pervasive? Why aren’t people more comfortable / confident in reasoning about them? Why do tables, lists, and calendars dominate corporate knowledge management?

Not bizarre after you realize that demographic is made up of craftspeople (digital illustrators, painters) who are misrepresenting their work as Art in the first place. Semantics can be so tricky. We call a lot of things Art and a lot of people artists.

If you wear a seatbelt and still get into a car accident, and die, is the seatbelt useless? These vaccines, like the flu shot, help protect you from ruinous outcomes (death, organ damage) It doesn’t magically transport you to an alternate universe where the virus doesn’t exist.

See also:

I’d open up the Chrome dev tools network tab to see why it was so slow And I’d see every modal, every view change, require a round trip to the server before anything could be shown on screen Sure the dev could’ve used a suspense fallback. But local-first would be mostly instant

Second this! Feels weird to use both Figma and Miro. All Figma needs is the ability to add “post-its” and a way to draw arrows between boxes easily and annotate on them.

There are plugins that let you GET data from any json file into a Figma file, but imagine if prototype interactions let you POST, PUT, or DELETE? You could make a functioning prototype where your chief design concern is how the pieces work together. Mocking UI + API together.

Wasted opportunity if they’re thinking of Arc as a consumer play. There’s so obviously a B2B gap that they’ve filled, and instead of doubling down on teams, I’m surprised that they’ve decided to give up the ghost. Someone will fill it for sure now they’ve shown the way

What a striking similarity between wealth management & product development: - they’re explore/exploit problems - practitioners divided between qual/quant, micro/macro - both have hobbyist & professional subcultures driven by divisive cult-like personalities - markets decide

Many Singaporeans who have the privilege to travel or work and live abroad also feel it’s a shame that SG has largely ditched its national identity for a more bland, globalist identity. It’s a shame because SG no longer feels like home.

Well, sure, but Apple in the 80s paid attention to usability & ergonomics of their products too. They didn’t have 161,000 employees then.

I think it’s a good thing that all of the world’s interaction design & UI engineering talent gets diverted from building CRUD apps, e-commerce apps, and FinTech apps that are better handled by natural language commands; and more towards “direct manipulation” creative apps.

The job title “designer” is such a misleading misnomer! I went into design thinking I’d get to design, turns out in most orgs, design decisions are largely made my Product Managers and most of the time, a non-trivial number of these decisions get overridden by engineers.

Just discovered this neat little feature in @raindrop_io. It seems like you can now highlight text on any webpage in Safari and it auto-bookmarks the page for you. Neat!

Just tried this. This is really, really cool. This, combined with Vercel’s @v0_dev, makes prototyping a functioning full stack web app possible now. Code generation is hit and miss if you get the LLM to just generate code from scratch without any real environment to execute generated code against. But v0 and postgres.new is guided and structured generation. In v0’s case, that’s real UI that’s rendered in the browser. In postgres.new’s case, real SQL is being generated and executed in an in-memory database. More and more, this means the messy part where you’re prototyping an app or feature is now far easier. A single designer or PM can do it, without the need for a cross-functional team. Granted, none of this is production ready, but eliminating the back-and-forth during prototyping shaves off so much time that it’s already revolutionary on its own.

Artists make money from royalties and primary sales, not commissions or work for hire. Craftspeople the opposite. You can see why machine learning models threaten their livelihoods. So it’s not so bizarre after all why they’re against it.

Not something easily visible to expats. SG govt policies have done a 180 in the past decade from being localist to being globalist, that has hurt a lot of folks for which the social contract is now broken.

Your notetaking app is an end-user database. Which means: Picking a notetaking app is like picking a database. Different ones have different paradigms and box you into modeling the world in different ways. Taking notes is like modeling data

People use Google Maps just fine on a daily basis. People are perfectly 👌 with inherent uncertainty in their daily commute. Why is it that in a work enviornment, people reach for the calendar as an organizing tool? It’s as if the seasons affect our yield, like we’re farmers! 👩‍🌾

Strong processes, loosely held? 😛

This UI, which maintains a 1:1 mapping between a document and an outline, is alluring. It reminds us that with the right affordances that play nicely with each other within the same system, we don’t have to choose between them in competing tools.

How powerful would it be if, instead of asking, “where is [work item] on the list?” or “when is [work item] going to be done?”, we asked: “What paths do we – or don’t we – travel, to get to where we want to go?” I’m talking about graph traversal as a planning tool.

I should also clarify (since a lot of people have pointed out that explore/exploit also includes an element of time) that this does not mean product development does not employ scheduling methods. Quite the contrary. You schedule efforts to negotiate the explore/exploit trade off

It really does feel like notetaking apps are bringing a (bespoke, artisanally-crafted) knife to a gun fight. It’s easier than ever before to produce and publish content. There’s just so much of it. Much, much harder to process it, understand it, apply it to practice.

For sure, although “large up-front research” can sometimes be unjustifiably vilified because it looks like “big up-front analysis/design”. Time taken to observe the market (synthesis) is valuable, spending a large amount of time analyzing your potential future moves, not so much

Not wanting to make something “useless” or something you wouldn’t be proud of, I suppose. Small projects tend to be underwhelming. Idealism/perfectionism can make this worse. Speaking from 1st hand experience lol.

Viewing Twitter as a social network is kind of reductive. It’s, rather, an *idea*, *opinion*, and *external media* graph stitched together by individuals (rather than by topics). It’s a multiplayer notetaking & bookmarking tool. “Microblogging”, it was called back in the day.

This video: piped.video/watch?v=2IW9L1uN… Then read learnwardleymapping.com/ Then read medium.com/wardleymaps And finally follow @swardley and @HiredThought

This tweet somehow manages to be both about the muskening and also has a LinkedIn vibe

I believe that the artist/maker who makes art/tools for themselves to enjoy/use, will always, always, make a better one than a mercenary paid to make something for someone else. “Better”, not by any technical standard, but by how it meets expectations.