In a way, the notification center on a mobile device has become the federation point for all communication apps, similar to how older desktop IM clients unified communication across multiple IM networks.

Federation gives us more collective control over what changes we accept, but that comes with an unacceptable inability to adapt.

I don’t think that what we are seeing is a “second wave”. I think we are seeing a seasonal effect. It’s important to keep some perspective. As I explained earlier in this article, cases are a very poor way to determine how active the virus is in the population. In Sweden, the number of tests being carried out is now eight times higher than in the spring. That is why we should instead be looking at hospitalizations, ICU admissions, and deaths.

I think it’ll probably be a few years before we know the full extent to which people were harmed by this. As an example, the hospital in Uppsala experienced 50% fewer admissions due to cardiac infarctions (“heart attacks”) during the peak period, while the hospitals in Stockholm experienced 40% fewer admissions. We know that people who have a cardiac infarction and don’t get emergency treatment have a significantly increased risk of dying in the immediate future, and also have a greater risk of developing long term complications such as heart failure.

The Swedish Public Health authority has never admitted that the goal of their chosen strategy is to reach herd immunity. However, from an epidemiological stand point, all strategies depend on reaching herd immunity in one way or another. A vaccination based strategy also builds on getting to herd immunity, it just chooses a different way to reach it. At some point in the relatively near future, every country on Earth will have developed herd immunity to covid, either by letting the disease spread until that point is reached, or by vaccinating enough people to reach that point.

The first is that the presence of surveillance means society cannot experiment with new things without fear of reprisal, and that means those experiments—if found to be inoffensive or even essential to society—cannot slowly become commonplace, moral, and then legal. If surveillance nips that process in the bud, change never happens. All social progress—from ending slavery to fighting for women’s rights—began as ideas that were, quite literally, dangerous to assert. Yet without the ability to safely develop, discuss, and eventually act on those assertions, our society would not have been able to further its democratic values in the way that it has. Consider the decades-long fight for gay rights around the world. Within our lifetimes we have made enormous strides to combat homophobia and increase acceptance of queer folks’ right to marry. Queer relationships slowly progressed from being viewed as immoral and illegal, to being viewed as somewhat moral and tolerated, to finally being accepted as moral and legal. In the end it was the public nature of those activities that eventually slayed the bigoted beast, but the ability to act in private was essential in the beginning for the early experimentation, community building, and organizing. Marijuana legalization is going through the same process: it’s currently sitting between somewhat moral, and—depending on the state or country in question—tolerated and legal. But, again, for this to have happened, someone decades ago had to try pot and realize that it wasn’t really harmful, either to themselves or to those around them. Then it had to become a counterculture, and finally a social and political movement. If pervasive surveillance meant that those early pot smokers would have been arrested for doing something illegal, the movement would have been squashed before inception. Of course the story is more complicated than that, but the ability for members of society to privately smoke weed was essential for putting it on the path to legalization. We don’t yet know which subversive ideas and illegal acts of today will become political causes and positive social change tomorrow, but they’re around. And they require privacy to germinate. Take away that privacy, and we’ll have a much harder time breaking down our inherited moral assumptions.

A standard example of a non-personalized nudge involves retirement planning. An employer could (i) leave it to employees to set-up their 401K plans and decide how much to save or (ii) set up the plans by default so that a predetermined amount is saved automatically and allow employees to make adjustments. Saving by default is an architected choice that relies on two facts: first, people often fail to set up a retirement plan, which is a social problem, and second, people tend to stick with default rules. Thus, by choosing option (ii), the choice architect nudges people to start with the better position for them and society.

For both the tailor-customer and doctor-patient examples, personal data is an input used to improve an output (dress, suit, medical treatment) such that the improvement directly serves the interests of the person whose information is being used.

Preserving user privacy is difficult when detectingmore nuanced forms of censorshipSome forms of softcensorship might involve intentional performance degrada-tion or content manipulation. Detecting this type of behav-ior would require comparing performance or content acrossgroups of users, but such a comparison also implies thateach user must reveal their browsing history.

If the EU is set to mandate encryption backdoors to enable law enforcement to pursue bad actors on social media, and at the same time intends to continue to pursue the platforms for alleged bad practices, then entrusting their diplomatic comms to those platforms, while forcing them to have the tools in place to break encryption as needed would seem a bad idea.

First, the recognition that sensitive information needs to be transmitted securely over instant messaging platforms plays into the hands of the privacy advocates who are against backdoors in the end-to-end encryption used on WhatsApp, Signal, Wickr, iMessage and others. The core argument from the privacy lobby is that a backdoor will almost certainly be exploited by bad actors. Clearly, the EU (and others) would not risk their own comms with such a vulnerability.

She was not his spouse; instead, she was the political officer of the Communist Party of China and she was embedded in the school. In fact, the university is full of political officers who operate behind the scenes but are there to keep party discipline. Today, companies like Nike, Google, Microsoft and others don’t need a communist party to impose their own totalitarian-like discipline upon workers.

Big corporations have been doing this for a while. In 1027, after Trump implemented the travel ban on people coming from countries that sponsor terrorism, Airbnb responded with an ad during the Super Bowl called “We Accept.” It featured people of different nationalities and declared, “We believe no matter who you are, where you’re from, who you love or who you worship, we all belong. The world is more beautiful the more you accept.”

In another interview, this time by John Laidler of the Harvard Gazette (March 2019), Zuboff expanded on this: I define surveillance capitalism as the unilateral claiming of private human experience as free raw material for translation into behavioral data. These data are then computed and packaged as prediction products and sold into behavioral futures markets

In order to maximize the effectiveness of your projections, figure out which things you want to be associated with in people’s heads and be excited about them. Nothing is more memorable than distinct excitement. Similarly, try to figure out the things that drive the whomever you’re talking to and incorporate them into your mental representation of that person.

When you’re implementing a bad plan yourself, instead of having a mentor bail you out by fixing it, a few really useful things happen:You learn many more details about why it was a bad idea. If someone else tells you your plan is bad, they’ll probably list the top two or three reasons. By actually following through, you’ll also get to learn reasons 4–1,217.You spend about 100x more time thinking about how you’ll avoid ever making that type of mistake again, i.e., digesting what you’ve learned and integrating it into your overall decision-making.By watching my mistakes and successes play out well or badly over the course of months, I was able to build much more detailed, precise models about what does and doesn’t matter for long-term codebase health. Eventually, that let me make architectural decisions with much more conviction.

The digital realm is overtaking and redefining everything familiar even before we have had a chance to ponder and decide.

But as long as the most important measure of success is short-term profit, doing things that help strengthen communities will fall by the wayside. Surveillance, which allows individually targeted advertising, will be prioritized over user privacy. Outrage, which drives engagement, will be prioritized over feelings of belonging. And corporate secrecy, which allows Facebook to evade both regulators and its users, will be prioritized over societal oversight.

Increased pressure on Facebook to manage propaganda and hate speech could easily lead to more surveillance. But there is pressure in the other direction as well, as users equate privacy with increased control over how they present themselves on the platform.

Facebook makes choices about what content is acceptable on its site. Those choices are controversial, implemented by thousands of low-paid workers quickly implementing unclear rules. These are tremendously hard problems without clear solutions. Even obvious rules like banning hateful words run into challenges when people try to legitimately discuss certain important topics.

At the same time, working through these principles is only the first step in building out a privacy-focused social platform. Beyond that, significant thought needs to go into all of the services we build on top of that foundation -- from how people do payments and financial transactions, to the role of businesses and advertising, to how we can offer a platform for other private services.

People want to be able to choose which service they use to communicate with people. However, today if you want to message people on Facebook you have to use Messenger, on Instagram you have to use Direct, and on WhatsApp you have to use WhatsApp. We want to give people a choice so they can reach their friends across these networks from whichever app they prefer.We plan to start by making it possible for you to send messages to your contacts using any of our services, and then to extend that interoperability to SMS too. Of course, this would be opt-in and you will be able to keep your accounts separate if you'd like.

An important part of the solution is to collect less personal data in the first place, which is the way WhatsApp was built from the outset.

As we build up large collections of messages and photos over time, they can become a liability as well as an asset. For example, many people who have been on Facebook for a long time have photos from when they were younger that could be embarrassing. But people also really love keeping a record of their lives.

We increasingly believe it's important to keep information around for shorter periods of time. People want to know that what they share won't come back to hurt them later, and reducing the length of time their information is stored and accessible will help.

I understand that many people don't think Facebook can or would even want to build this kind of privacy-focused platform -- because frankly we don't currently have a strong reputation for building privacy protective services, and we've historically focused on tools for more open sharing.

But now, with all the ways people also want to interact privately, there's also an opportunity to build a simpler platform that's focused on privacy first.

Today we already see that private messaging, ephemeral stories, and small groups are by far the fastest growing areas of online communication.

As I think about the future of the internet, I believe a privacy-focused communications platform will become even more important than today's open platforms. Privacy gives people the freedom to be themselves and connect more naturally, which is why we build social networks.

AI is not analogous to the big science projects of the previous century that brought us the atom bomb and the moon landing. AI is a science that can be conducted by many different groups with a variety of different resources, making it closer to computer design than the space race or nuclear competition. It doesn’t take a massive government-funded lab for AI research, nor the secrecy of the Manhattan Project. The research conducted in the open science literature will trump research done in secret because of the benefits of collaboration and the free exchange of ideas.

Now let me get back to your question. The FBI presents its conflict with Apple over locked phones as a case as of privacy versus security. Yes, smartphones carry a lot of personal data—photos, texts, email, and the like. But they also carry business and account information; keeping that secure is really important. The problem is that if you make it easier for law enforcement to access a locked device, you also make it easier for a bad actor—a criminal, a hacker, a determined nation-state—to do so as well. And that's why this is a security vs. security issue.

The FBI — along with many other law enforcement and surveillance agents — insists that it is possible to make crypto that will protect our devices, transactions, data, communications and lives, but which will fail catastrophically whenever a cop needs it to. When technologists explain that this isn't a thing, the FBI insists that they just aren't nerding hard enough.

Barr makes the point that this is about “consumer cybersecurity” and not “nuclear launch codes.” This is true, but it ignores the huge amount of national security-related communications between those two poles. The same consumer communications and computing devices are used by our lawmakers, CEOs, legislators, law enforcement officers, nuclear power plant operators, election officials and so on. There’s no longer a difference between consumer tech and government tech—it’s all the same tech.

The answers for law enforcement, social networks, and medical data won’t be the same. As we move toward greater surveillance, we need to figure out how to get the best of both: how to design systems that make use of our data collectively to benefit society as a whole, while at the same time protecting people individually.

COVID-19 adds a new urgency to the debate and brings in new actors such as public health authorities and the medical sector. It’s not just about smartphone apps tracing contacts with infected people that are currently being rolled out by corporations and governments around the world. The medical community will seize the pandemic to boost its case for accessing detailed health data to perform all sorts of research studies. Public health authorities will push for more surveillance in order to get early warning of future pandemics. It’s the same trade-off. Individually, the data is very intimate. But collectively, it has enormous value to us all.

There’s been a fundamental battle in Western societies about the use of personal data, one that pits the individual’s right to privacy against the value of that data to all of us collectively. Until now, most of that discussion has focused on surveillance capitalism. For example, Google Maps shows us real-time traffic, but it does so by collecting location data from everyone using the service.

Too many wrongly characterize the debate as “security versus privacy.” The real choice is liberty versus control. Tyranny, whether it arises under threat of foreign physical attack or under constant domestic authoritative scrutiny, is still tyranny. Liberty requires security without intrusion, security plus privacy. Widespread police surveillance is the very definition of a police state. And that’s why we should champion privacy even when we have nothing to hide.

A future in which privacy would face constant assault was so alien to the framers of the Constitution that it never occurred to them to call out privacy as an explicit right. Privacy was inherent to the nobility of their being and their cause.

We do nothing wrong when we make love or go to the bathroom. We are not deliberately hiding anything when we seek out private places for reflection or conversation. We keep private journals, sing in the privacy of the shower, and write letters to secret lovers and then burn them. Privacy is a basic human need.

Privacy protects us from abuses by those in power, even if we’re doing nothing wrong at the time of surveillance.

My problem with quips like these — as right as they are — is that they accept the premise that privacy is about hiding a wrong. It’s not. Privacy is an inherent human right, and a requirement for maintaining the human condition with dignity and respect.

As society scales up, gossip becomes ineffective. Rumors don’t spread easily from village to village, so I can get away with violating norms when I venture out and deal with strangers.

Someone who is skilled at acquiring and spreading information can have high status in the tribe. As receivers of gossip, we are excited to be “in on the secret.”

Telling lies may eventually get you in trouble, but not necessarily. If the victims of a false rumor are unable to fight back effectively, people who engage in false gossip may be successful.

To escape from the chaos, we will need new norms of behavior that incline us away from gossip.

The result is that we are living through a period of chaos. Symptoms include conspiracy theories, information bubbles, cancel culture, President Trump’s tweets, and widespread institutional decay and dysfunction.

We have increased the power of gossip-mongers and correspondingly reduced the power of elite institutions of the 20th century, including politicians, mainstream media, and scientists.

Our ISS technology changes this. It makes it possible to gossip effectively at large scale. This in turn has revived our propensity to rely on gossip. Beliefs spread without being tested for truth.

But gossip is not the search for truth. It is a search for approval by attacking the perceived flaws of others.

Large societies need other enforcement mechanisms: government, religion, written codes.

As a social enforcement mechanism, gossip does not scale.

Human evolution produced gossip. Cultural anthropology sees gossip as an informal way of enforcing group norms. It is effective in small groups.

Declarative programming is an enabler of abstraction. Imperative programming is an inhibitor of abstraction. Declarative programming allows you to say “I want this and I don’t care how I get it” while imperative programming requires you to define each and every step.

By 2005 blogs had crashed the cultural gates. China’s editors, station directors, and pub-lishers had always acted as cultural “gatekeepers:” deciding who could and couldn’t becomeknown through publication, TV and film appearances, and musical performances. In a majorcultural power-shift, pop cultural icons could emerge through blogs, forums, chatrooms, andpersonal websites, completely outside of the government approved cultural structures.But while Communist Party propaganda department had lost control over China’s cul-ture, in the realm of politics the gates and walls are constantly being rebuilt, upgraded, andreinforced. It would be impossible for a dissident political leader to rise to popularity in thesame way that Mu Zimei rose to stardom.

This situation is reinforced by recent survey results—surprisingto many Westerners—showing that most urban Chinese Internet users actually trust domesticsources of news and information more than they trust the information found on foreign newswebsites (Guo et al.2005, pp. 66–67).

While the Chinese government has supportedthe development of the Internet as a tool for business, entertainment, education, and infor-mation exchange, it has succeeded in preventing people from using the Internet to organizeany kind of viable political opposition.

They are determined to prevent the Internet from serving as a tool for “colorrevolution” in the way that online media and communication tools empowered activists inUkraine and Lebanon. Thus in 2005 the Chinese government updated its regulations control-ling online news and information, and aggressively leaned on organizations hosting onlinechatrooms and blogs to stop the spread of online discussions about recent local governmentcrackdowns against farmer protests in the Chinese countryside.

Though government statements emphasize anti-pornography crackdowns, ONI found the primary focus of China's filtering system to be on political content. Public security organs and internet service providers employ thousands of people – nationwide, at multiple levels – as monitors and censors. Their job is to monitor everything posted online by ordinary Chinese people and to delete objectionable content.

Most people seem to follow one of two strategies - and these strategies come under the umbrella of tree-traversal algorithms in computer science.

Here's one quick way to test if your application has properly segregated itself between the Model, View, and Controller roles: is your app skinnable? My experience is that designers don't understand loops or any kind of state. They do understand templates with holes in them. Everybody understands mail merge. And if you say, "Apply the bold template to this hole," they kind of get that, too. So separating model and view addresses this very important practical problem of how to have designers work with coders. The other problem is there is no way to do multiple site skins properly if you don't have proper separation of concerns. If you are doing code generation or sites with different skins on them, there is no way to properly make a new skin by simply copying and pasting the old skin and changing it. If you have the view and the logic together, when you make a copy of the view you copy the logic as well. That breaks one of our primary rules as developers: have only one place to change anything.

In 1972 David L. Parnas published a classic paper entitled On the Criteria To Be Used in Decomposing Systems into Modules. It appeared in the December issue of the Communications of the ACM, Volume 15, Number 12. In this paper, Parnas compared two different strategies for decomposing and separating the logic in a simple algorithm. The paper is fascinating reading, and I strongly urge you to study it. His conclusion, in part, is as follows: “We have tried to demonstrate by these examples that it is almost always incorrect to begin the decomposition of a system into modules on the basis of a flowchart. We propose instead that one begins with a list of difficult design decisions or design decisions which are likely to change. Each module is then designed to hide such a decision from the others.”

"Let me try to explain to you, what to my taste is characteristic for all intelligent thinking. It is, that one is willing to study in depth an aspect of one's subject matter in isolation for the sake of its own consistency, all the time knowing that one is occupying oneself only with one of the aspects. We know that a program must be correct and we can study it from that viewpoint only; we also know that it should be efficient and we can study its efficiency on another day, so to speak. In another mood we may ask ourselves whether, and if so: why, the program is desirable. But nothing is gained —on the contrary!— by tackling these various aspects simultaneously. It is what I sometimes have called "the separation of concerns", which, even if not perfectly possible, is yet the only available technique for effective ordering of one's thoughts, that I know of. This is what I mean by "focussing one's attention upon some aspect": it does not mean ignoring the other aspects, it is just doing justice to the fact that from this aspect's point of view, the other is irrelevant. It is being one- and multiple-track minded simultaneously.

Domain-driven design separates the model layer “M” of MVC into an application, domain and infrastructure layer. The infrastructure layer is used to retrieve and store data. The domain layer is where the business knowledge or expertise is. The application layer is responsible for coordinating the infrastructure and domain layers to make a useful application. Typically, it would use the infrastructure to obtain the data, consult the domain to see what should be done, and then use the infrastructure again to achieve the results.

Q7. What are controlled components?In HTML, form elements such as <input>, <textarea>, and <select> typically maintain their own state and update it based on user input. When a user submits a form the values from the aforementioned elements are sent with the form. With React it works differently. The component containing the form will keep track of the value of the input in it's state and will re-render the component each time the callback function e.g. onChange is fired as the state will be updated. A form element whose value is controlled by React in this way is called a "controlled component".With a controlled component, every state mutation will have an associated handler function. This makes it straightforward to modify or validate user input.

React creates a virtual DOM. When state changes in a component it firstly runs a “diffing” algorithm, which identifies what has changed in the virtual DOM. The second step is reconciliation, where it updates the DOM with the results of diff.The HTML DOM is always tree-structured — which is allowed by the structure of HTML document. The DOM trees are huge nowadays because of large apps. Since we are more and more pushed towards dynamic web apps (Single Page Applications — SPAs), we need to modify the DOM tree incessantly and a lot. And this is a real performance and development pain.The Virtual DOM is an abstraction of the HTML DOM. It is lightweight and detached from the browser-specific implementation details. It is not invented by React but it uses it and provides it for free. ReactElements lives in the virtual DOM. They make the basic nodes here. Once we defined the elements, ReactElements can be render into the "real" DOM.Whenever a ReactComponent is changing the state, diff algorithm in React runs and identifies what has changed. And then it updates the DOM with the results of diff. The point is - it’s done faster than it would be in the regular DOM.

The state is a data structure that starts with a default value when a Component mounts. It may be mutated across time, mostly as a result of user events.

Props (short for properties) are a Component’s configuration. Props are how components talk to each other. They are received from above component and immutable as far as the Component receiving them is concerned. A Component cannot change its props, but it is responsible for putting together the props of its child Components. Props do not have to just be data — callback functions may be passed in as props.

Q5. What are the differences between a class component and functional component?Class components allows us to use additional features such as local state and lifecycle hooks. Also, to enable our component to have direct access to our store and thus holds state.When our component just receives props and renders them to the page, this is a ‘stateless component’, for which a pure function can be used. These are also called dumb components or presentational components.

Q2. What is JSX?JSX is a syntax extension to JavaScript and comes with the full power of JavaScript. JSX produces React “elements”. You can embed any JavaScript expression in JSX by wrapping it in curly braces. After compilation, JSX expressions become regular JavaScript objects. This means that you can use JSX inside of if statements and for loops, assign it to variables, accept it as arguments, and return it from functions. Eventhough React does not require JSX, it is the recommended way of describing our UI in React app.

Taking this one step further, a big part of the friction that Zoom removed was that you don’t need an account, an app or a social graph to use it: Zoom made network effects irrelevant. But, that means Zoom doesn’t have those network effects either. It grew by removing defensibility.

There’s lots of bundling and unbundling coming, as always. Everything will be ‘video’ and then it will disappear inside.

Part of the founding legend of Dropbox is that Drew Houston told people what he wanted to do, and everyone said ‘there are hundreds of these already’ and he replied ‘yes, but which one do you use?’ That’s what Zoom did - video calls are nothing new, but Zoom solved a lot of the small pieces of friction that made it fiddly to get into a call.  

To cultivate an idea meritocracy, they developed an app called a “dot collector” which enables all employees to rate each other along many different dimensions, ranging from “knowledgeability” to communication style. Over time, the app builds up a picture of each employee’s “believability” on different issues. This enables Bridgewater to understand where expertise lies within the company in addition to the hierarchical authority easily understood on an org chart.

The chaos manager is concerned with the credibility of the organization and ensures that positional authority is aligned with personal authority.  That the people in leadership are the ones people want to follow.  While the Marine Corps has a clear position hierarchy, they have a deep understanding of this idea.  Official authority is a function of rank and position and is bestowed by organization and by law. Personal authority is a function of personal influence and derives from factors such as experience, reputation, skill, character, and personal example. It is bestowed by the other members of the organization.…Official authority provides the power to act but is rarely enough; most effective commanders also possess a high degree of personal authority

Amazon pushes teams to escalate one-way door decisions – those that can’t be reversed and may have long-term consequences.  However, with “two-way” decisions, managers are coached to make these decisions themselves.

“Working backwards”  from customer needs can be contrasted with a “skills-forward” approach where existing skills and competencies are used to drive business opportunities. The skills-forward approach says, “We are really good at X. What else can we do with X?” That’s a useful and rewarding business approach. However, if used exclusively, the company employing it will never be driven to develop fresh skills.

However, I quickly realized the problem.  Kotter’s approach puts the senior executive at the center of the story and the leader’s task is to force a change on a resistant organization.  To him, the business leader “defines what the future should look like, aligns people with that vision, and inspires them to make it happen despite the obstacles” Chaos theory, in contrast, removes the senior executive from the center of the story and puts the system at the center.  That is exciting for people who enjoy thinking about complex systems, but isn’t likely to be profitable to a consulting firm which sells projects to senior executives.

Historically, design has been opaque as a business unit due to the logistical and technical difficulties of making the design process legible to others. But as these hurdles are increasingly solved by companies like Figma, we’re seeing teams navigate how to best integrate design with the rest of a company’s processes. This should be no surprise, as we have seen the same arc play out in engineering over the last few decades.

Figma is browser-first, which was made possible (and more importantly performant) by their understanding and usage of new technologies like WebGL, Operational Transforms, and CRDTs. From a user’s perspective, there are no files and no syncing that needs to be done with others editing a design. The actual *experience* of designing in Figma is native to the internet. Even today, competitors often talk about cloud, but are torn over how *much* of the experience to port over to the internet. Hint: “all of it” is the correct answer that they all eventually will converge on.

In 2014 as I helped diligence Figma (disclaimer: I worked on Greylock’s investment in Figma, but I don’t have a personal stake in Figma. sadly.), I used to sit with designers at startups and watch them work. The top right corner of their screens were always a nonstop cycle of Dropbox notifications. Because design teams saved all their files in a shared folder like Dropbox, every time a coworker made a revision they would get a notification. And often there were complex naming conventions to make sure that people were using the right versions. Figma solved this problem. Designs in Figma are not just stored in the cloud; they are edited in the cloud, too. This means that Figma users are always working on the same design. With Dropbox, this isn’t true. The files may be stored in the cloud, but the editing happens locally—imagine the difference between sharing Word files in Dropbox vs. editing in Google Docs.

Design appears to be inflecting in the direction of engineering. Figma is in pole position to drive this evolution. As a tool, it makes designers both more efficient and more collaborative by breaking down the walls between design and the other teams they work with.

As disciplines evolve, they figure out the social norms needed to operate better, build tools that can be shared across the industry, and invent abstractions that allow offloading more and more of the workload. They learn how to collaborate better, not just with each other but with all the other functions as well. Disciplines are not an end to themselves; the degree to which they contribute to the larger organizations and ecosystems they are part of is the final measure of their progress.

Engineering is almost unparalleled in the rate at which it commoditizes itself and pushes the frontier of progress out. The best practices in frameworks, languages, and infrastructure are always rapidly—and sometimes tumultuously—evolving. What used to take entire teams to build before, requires fewer and fewer people every year.

Platforms are needed most when the diversity and scale of use cases is larger than can be built—or often even understood—by the company.

The real power of plugins, however, is in making them publicly available across the ecosystem. Plugins are collective progress available to all users.

In many ways Figma’s Communities are a reflection of Github’s philosophy and intent, but built with design in mind. Duplicate a shared design, and a copy is instantly saved to your workspace and ready to be edited.

This impacts monetization and purchasing at companies. Paying for a new design tool because it has new features for designers may not be a top priority. But if product managers, engineers, or even the CEO herself think it matters for the business as a whole—that has much higher priority and pricing leverage.

By bringing both designers and non-designers alike into Figma, they create a cross-side network effect. In a direct network effect, a homogenous group gets more value from a product as more of them join. In contrast, a cross-side network effect involves two (or more) distinct groups that grow in size and value as the other group does, too. Figma’s cross-side network effect between designers and non-designers is one of the primary and under-appreciated sources of their compounding success over the last few years. As more designers use Figma, they pull in the non-designers they work with. Similarly, as these non-designers use Figma, they encourage the other designers they work with to use Figma. It’s a virtuous circle and a powerful compounding loop.

Much of Figma’s current success is driven by its ability to spread within companies. Figma becomes more useful as more people within a company use it, driving advantaged speed and scale of penetration within companies. Figma was quick to recognize that the constraints on design at companies is often not a problem of pixels, but of people. Many of Figma’s competitors are great tools for designers. But that’s who they are for—designers. Figma is a tool for teams to design. Not for designers alone.

Tightening the feedback loop of collaboration allows for non-linear returns on the process. Design can be drafted simultaneously with the product, allowing feedback to flow in both directions throughout the process. Aligning the assets used by design and engineering allow more seamless handoffs, and allows for more lossless and iterative exchange.

Increasingly our tools must understand and align with how we collaborate. This was less important when collaboration was logistically difficult and prohibitively costly, but as collaboration becomes easier its importance has risen. People’s work is less siloed—and their tools must reflect this.

Our understanding of building platforms and sequencing towards them is still nascent. From how to shift the allocation of resources between a company’s core business and its potential future expansions to how to structure a platform to catalyze its growth and more. Until the playbook is well understood, it is more art than science. There are many decisions to make, including which layers should be centralized, whether the ecosystem should be driven by open source contributors or profit-seeking enterprises, how broad in scope to allow the ecosystem to grow and where to not let it grow, and more.

Building for everyone in the design process and not just designers is also the foundation of Figma’s core loop, which drives their growth and compounding scale. That network effect is made possible by Figma’s key early choices like:  Architecting Figma to be truly browser-first, instead of just having storage be in the cloudTheir head start in new technologies like WebGL and CRDTs that made this browser-first approach possibleFocusing on a product purpose built for those designing vector based digital products

The core insight of Figma is that design is larger than just designers. Design is all of the conversations between designers and PMs about what to build. It is the mocks and prototypes and the feedback on them. It is the handoff of specs and assets to engineers and how easy it is for them to implement them.

In a pure cloud world, this atomic unit of documents seems increasingly archaic. Documents are more a constraint of a pre-cloud world. And once you assume storing them online is table stakes, the question becomes where is actual collaboration happening that then leads people to wherever they need to do work.

The arc of collaboration is long and it bends in the direction of functional workflows.

Messaging, it turned out, appears to be a better center of gravity than documents. And while Dropbox (barring significant traction in its new products) seems to be fading in its centrality, what’s striking is that Slack’s victory seems hollow as well. If anything we’ve seen even *more* new companies building towards owning parts of these workflows and getting traction.

When Slack first started growing, there were many debates over which company would own collaboration, Slack or Dropbox. Dropbox proponents argued that Dropbox already managed all the actual records of a company, and so would be the center of gravity. Slack partisans argued that Dropbox was a transitory product, and eventually companies would stop caring about individual files, and messaging would be the more important live heartbeat of a company.

Beyond its Slack-like functionality, Discord has functionality like a social graph, seeing what games your friends are playing, voice chat, etc. These have been misunderstood by the market. They aren’t random small features. They are the backbone of a central nervous system. Active users of Discord have it on all the time, even when they are not playing games. It’s a passive way to have presence with your friends. And when your friends start playing games it makes it easy to with one click go join them in the game. Bringing your actual social graph across all games. Finally, voice chat makes it possible to talk with your friends across all games, even when you are playing the game. Like when working in a google doc, having to switch out of your game to message is a negative experience. Instead Discord adds functionality to your games even while you are focused solely on them. We will see more companies understand and begin to work on this area.

Slack ironically is more similar to Dropbox than expected. The more time goes by the more it looks like exception handling being needed ubiquitously is a transitory product as we switch off of documents. After all, like Dropbox, Slack makes the most sense as a global communication channel when the workflows themselves don’t have communication and collaboration baked in natively. For documents this is true, but increasingly for modern apps this is false.

As a company’s processes mature and the apps they use get more sophisticated, we expect to see the need to go to Slack for exception handling *decrease* over time.

Slack serves three functions: Else statement. Slack is the exception handler, when specific productivity apps don’t have a way to handle something. This should decrease in usefulness, as the apps build in handling of these use cases, and the companies build up internal processes.Watercooler. Slack is a social hub for co-workers. This is very important, and full of gifs.Meta-coordination. Slack is the best place for meta-levels of strategy and coordination that don’t have specific productivity apps. This is really a type of ‘else statement’, but one that could persist for a while in unstructured format.

It’s not that Slack is too distracting and killing individual productivity. It’s that your company’s processes are so dysfunctional you need Slack to be distracting and killing individual productivity.

The dream of Slack is that they become the central nervous system for all of a company’s employees and apps. This is the view of a clean *separation* of productivity and collaboration. Have all your apps for productivity and then have a single app for coordinating everyone, with your apps also feeding notifications into this system. In this way, Slack would become a star. With every app revolving around it. Employees would work out of Slack, periodically moving to whichever app they were needed in, before returning to Slack. But productivity *isn’t* separate from collaboration. They are the two parts of the same loop of producing work. And if anything collaboration is in *service* of team productivity.

And core Dropbox is not a solution to this. People store their documents in it. But they had to use email and other messaging apps to tell their co-workers which document to check out and what they needed help with. Dropbox understands this concern. It’s what’s driven their numerous forays into owning the workflows and communication channels themselves. With Carousel, Mailbox, and their new desktop apps all working to own that. However, there are constraints to owning the workflow when your fundamental atomic unit is documents. And they never quite owned the communication channels.

Suddenly, the constraint on work became much more about the speed and lossiness of collaboration. Which remained remarkably analog. The friction of getting people your document, much less keeping correct versioning was non-trivial.

Digital work has significantly faster feedback loops for productivity. Software, quite simply, can produce and iterate new things at a daily if not hourly or minute basis.

As the ecosystem of specialized SaaS apps and workflows continues to mature, messaging becomes a place of last resort. When things are running smoothly, work happens in the apps built to produce them. And collaboration happens within them. Going to slack is increasingly a channel of last resort, for when there’s no established workflow of what to do. And as these functional apps evolve, there are fewer and fewer exceptions that need Slack. In fact, a sign of a maturing company is one that progressively removes the need to use Slack for more and more situations.

There’s just so much noise small businesses tend to ignore. But in Indonesia, that isn’t the case…yet. The software landscape there is similar to the 1990s in the US. It’s harder to piggyback off of existing software infrastructure — whether it’s payments or platforms — but there’s also a lot of obvious opportunity in software that no one is going after. The same could be said about investing elsewhere in Southeast Asia or in LatAm or Africa. There are fewer startups to compete with for attention, and it’s less of a marketing game than building a software company in the US.

As economist Carlotta Perez describes, we are now in the Deployment Phase of the internet in the US — meaning, we are in-process of exhausting all use cases for internet technologies in the US. What has traditionally happened at the end of a technology phase is oversaturation of investment dollars chasing smaller returns. Valuations go up, returns go down, and investors lose their money. (Sound familiar?) On a company level, what this means is, if not careful, a lot of companies will end up wasting marketing dollars in this type of landscape. Companies in the 2020s, unlike in the 1990s, need to really be performance-marketing driven in order to compete. The end of last year certainly showed us many examples of well-funded companies that could not make the unit economics work. The software industry has become a marketing game.

There are certainly exceptions but if we are talking strictly about software, (not hardware, not drug discovery, not synthetic bio, etc) you’d be hard pressed to find a company where winning does not require a solid marketing and/or sales game. This is very different from the 1990s. Having a marketing skillset and mindset is what you need to win in 2020 in the US software market.

As such, while many VCs are still fixated on finding unique technology in software and chasing companies that will ultimately be the sole winner, I’d contend that these two strategies — while successful in the 90s and early 00s — largely no longer work.

Put another way, a lot of the “low hanging fruit” in the US software market is now gone. Software in the US generally works. And new opportunities get swept up with would-be competitors immediately. If the 90s was about thinking through your build, the 2020s is about thinking through marketing & distribution.

Furthermore, incumbents who generally do a good job, often manage to continue reigning. According to Brad Gerstner, CEO of Altimeter Capital, who recently did a podcast on Invest Like The Best, large tech companies have managed to take even more market share than 10 years ago. Some people may argue this is because the large tech companies have improved their products over time to stay ahead due to their increased collection of data and better algorithms that feed on that data over time. That may be true for some companies but not all. This also applies to other products that have not made significant strides in their technology — Craigslist, Salesforce CRM, Turbotax, Quickbooks to name a few. Even Google Search which arguably had a better product in the 1990s compared to its peers is about on par with alternative search engines today, but 90% of people worldwide still use Google. Old habits die hard, and distribution matters more than ever if you are just starting a business. It’s hard to topple incumbents who have strong distribution and already large audiences — even if you can build a much better product.

For most software businesses in the US, the problem isn’t technical knowledge anymore. The problem is getting a wedge into distribution — also known as marketing.

This has led to a flurry of many applications being built online – often with multiple teams building the same thing. It is not uncommon to run into 50 different founding teams all trying to build a marketplace for gym trainers. Or 300 founding teams trying re-invent marketing automation.

GPT-2 is instantiated on giant supercomputers; it’s a safe bet they could calculate the square root of infinity in a picosecond. But it counts more or less the same way as a two-year old. GPT-2 isn’t doing math. It’s doing the ridiculous “create a universe from first principles and let it do the math” thing that humans do in their heads.

A machine learning researcher writes me in response to yesterday’s post, saying: I still think GPT-2 is a brute-force statistical pattern matcher which blends up the internet and gives you back a slightly unappetizing slurry of it when asked.

But this should be a wake-up call to people who think AGI is impossible, or totally unrelated to current work, or couldn’t happen by accident. In the context of performing their expected tasks, AIs already pick up other abilities that nobody expected them to learn. Sometimes they will pick up abilities they seemingly shouldn’t have been able to learn, like English-to-French translation without any French texts in their training corpus. Sometimes they will use those abilities unexpectedly in the course of doing other things. All that stuff you hear about “AIs can only do one thing” or “AIs only learn what you program them to learn” or “Nobody has any idea what an AGI would even look like” are now obsolete.

Wittgenstein writes: “The limits of my language mean the limits of my world”. Maybe he was trying to make a restrictive statement, one about how we can’t know the world beyond our language. But the reverse is also true; language and the world have the same boundaries. Learn language really well, and you understand reality. God is One, and His Name is One, and God is One with His Name. “Become good at predicting language” sounds like the same sort of innocent task as “become good at Go” or “become good at Starcraft”. But learning about language involves learning about reality, and prediction is the golden key. “Become good at predicting language” turns out to be a blank check, a license to learn every pattern it can.

Imagine you prompted the model with “What is one plus one?” I actually don’t know how it would do on this problem. I’m guessing it would answer “two”, just because the question probably appeared a bunch of times in its training data. Now imagine you prompted it with “What is four thousand and eight plus two thousand and six?” or some other long problem that probably didn’t occur exactly in its training data. I predict it would fail, because this model can’t count past five without making mistakes. But I imagine a very similar program, given a thousand times more training data and computational resources, would succeed. It would notice a pattern in sentences including the word “plus” or otherwise describing sums of numbers, it would figure out that pattern, and it would end up able to do simple math. I don’t think this is too much of a stretch given that GPT-2 learned to count to five and acronymize words and so on.

Again, GPT-2 isn’t good at summarizing. It’s just surprising it can do it at all; it was never designed to learn this skill. All it was designed to do was predict what words came after other words. But there were some naturally-occurring examples of summaries in the training set, so in order to predict what words would come after the words tl;dr, it had to learn what a summary was and how to write one.

A very careless plagiarist takes someone else’s work and copies it verbatim: “The mitochondria is the powerhouse of the cell”. A more careful plagiarist takes the work and changes a few words around: “The mitochondria is the energy dynamo of the cell”. A plagiarist who is more careful still changes the entire sentence structure: “In cells, mitochondria are the energy dynamos”. The most careful plagiarists change everything except the underlying concept, which they grasp at so deep a level that they can put it in whatever words they want – at which point it is no longer called plagiarism.

It might be instructive to think about what it would take to create a program which has a model of eighth grade science sufficient to understand and answer questions about hundreds of different things like “growth is driven by cell division”, and “What can magnets be used for” that wasn’t NLP led. It would be a nightmare of many different (probably handcrafted) models. Speaking somewhat loosely, language allows for intellectual capacities to be greatly compressed. From this point of view, it shouldn’t be surprising that some of the first signs of really broad capacity- common sense reasoning, wide ranging problem solving etc., have been found in language based programs- words and their relationships are just a vastly more efficient way of representing knowledge than the alternatives.

However we should not forget that the relationships between words are isomorphic to the relations between things- that isomorphism is why language works. This is to say the patterns in language use mirror the patterns of how things are(1). Models are transitive- if x models y, and y models z, then x models z. The upshot of these facts are that if you have a really good statistical model of how words relate to each other, that model is also implicitly a model of the world.

With more and more productivity apps creating their own messaging systems, users suddenly face a new problem: Multiple inboxes. You now have to check notifications in Github, Trello, Google Docs and half a dozen (if not more) other tools in your productivity stack.

With a strong enough NLP engine behind the command line interface, the possibilities become endless: Add that New York Times article to your Pocket queue or send it directly to your Kindle to read it later Re-assign Jira tickets directly from Superhuman or send them to your to-do listPay invoices or send money to a friend

Bottom line: Blockchain can help a bit with voting, but it’s not doing the most important part of the work. It doesn’t help tally secret ballots in a publicly verifiable way. It doesn’t provide individual verifiability that a ballot was correctly encoded. And it’s not useful for voting eligibility, since that’s all about human authentication and a centrally produced voter list. At best, in voting, Blockchain can be a ledger that helps us track the voting metadata.

Then there’s the need to check voter eligibility, a critical piece of global verifiability. No matter what technology we use, we need a clear list of eligible voters, and each voter should get to vote only once. Ultimately, the list of eligible voters is set in a centralized way: it’s produced by the State. There’s nothing distributed about voter eligibility. Even when there is federation / delegation to individual counties, like in the US, there is a centralized effort to cross-check that a voter isn’t registered in multiple counties.

In a typical election setting with secret ballots, we need: enforced secrecy: a way for each voter to cast a ballot secretly and no way to prove how they voted (lest they be unduly influenced) individual verifiability: a way for each voter to gain confidence that their own vote was correctly recorded and counted. global verifiability: a way for everyone to gain confidence that all votes were correctly counted and that only eligible voters cast a ballot.

Blockchain isn’t just a distributed database, it’s a very specific kind of distributed database where the database maintainers aren’t authenticated: anyone can be a blockchain maintainer without revealing who they are or having any kind of privileged relationship with other maintainers. the set of maintainers changes over time. New maintainers come in, existing maintainers leave, without central planning or predictability. The maintainers of the Bitcoin blockchain 5 years ago are very different from the maintainers today.

Ohhh, never thought of this hypothesis: that the act of getting drunk together might be a social technology that helps us verify the trustworthiness of others by inhibiting their higher cognitive functions and thus making it harder to consciously fake things.

As a final practical maxim, relative to these habits of the will, we may, then, offer something like this: Keep the faculty of effort alive in you by a little gratuitous exercise every day. That is, be systematically ascetic or heroic in little unnecessary points, do every day or two something for no other reason than that you would rather not do it, so that when the hour of dire need draws nigh, it may find you not unnerved.

Seize the very first possible opportunity to act on every resolution you make, and on every emotional prompting you may experience in the direction of the habits you aspire to gain. It is not in the moment of their forming, but in the moment of their producing motor effects, that resolves and aspirations communicate the new ‘set’ to the brain.

The more of the details of our daily life we can hand over to the effortless custody of automatism, the more our higher powers of mind will be set free for their own proper work.

Consider texts like the Bible and the Analects of Confucius. People integrate ideas from those books into their lives over time—but not because authors designed them that way. Those books work because they’re surrounded by rich cultural activity. Weekly sermons and communities of practice keep ideas fresh in readers’ minds and facilitate ongoing connections to lived experiences.

By now you might be starting to think about businesses and communities could benefit from paid social networks.

Just as built-in social networks are a moat for information products, customized tooling is a moat for social networks.

Today’s existing tools will continue to be sufficient for some communities, and Discord and Slack’s robust bot APIs are capable of solving some community needs. But fundamentally, they are still based on chat, and chat simply isn’t the right core user experience for many other communities. Unique functionality and bespoke interfaces provide distinct advantages that off-the-shelf tooling can never achieve.

If you believe there's nothing true that you can't say, then anyone who gets in trouble for something they say must deserve it.

It doesn't seem to conventional-minded people that they're conventional-minded. It just seems to them that they're right. Indeed, they tend to be particularly sure of it.

I do force myself to publish a Medley every week since they require me to keep reading.

Imagine a large population of people living, seeing, learning, doing and generally going about their lives. As they do so, they accumulate beliefs. Depending on how smart they are, they also compress beliefs via abstraction, metaphor, subconscious pattern-recognition circuits, muscle memory, ritual, making and consuming art, going p-value fishing, exploring tantric sex, generating irreproducible peer-reviewed Science! and so on.

Academics gain prestige by publishing novel stuff. This gives them a warped perspective on what is valuable. You can’t publish a paper that would summarize five other papers and argue that these papers are undervalued in a top journal but in the real world the value of doing that might be very high. The mechanisms of discovery are broken in academia.

The world is currently overflowing with hard problems, so it is overflowing with grift as well.

Grifts often rely on narrative vacuums. When the real story is too complicated or boring or requires numbers and graphs to understand, people reach for the simpler story. Grifters supply it.

A remarkable phenomenon commented on in the Moynihan report of thirty years ago goes unnoticed in The Bell Curve--the prevalence of females among blacks who score high on mental tests. Others who have done studies of high- IQ blacks have found several times as many females as males above the 120 IQ level. Since black males and black females have the same genetic inheritance, this substantial disparity must have some other roots, especially since it is not found in studies of high-IQ individuals in the general society, such as the famous Terman studies, which followed high-IQ children into adulthood and later life. If IQ differences of this magnitude can occur with no genetic difference at all, then it is more than mere speculation to say that some unusual environmental effects must be at work among blacks. However, these environmental effects need not be limited to blacks, for other low-IQ groups of European or other ancestries have likewise tended to have females over-represented among their higher scorers, even though the Terman studies of the general population found no such patterns. One possibility is that females are more resistant to bad environmental conditions, as some other studies suggest. In any event, large sexual disparities in high-IQ individuals where there are no genetic or socioeconomic differences present a challenge to both the Herrnstein- Murray thesis and most of their critics.

Strangely, Herrnstein and Murray refer to "folklore" that "Jews and other immigrant groups were thought to be below average in intelligence. " It was neither folklore nor anything as subjective as thoughts. It was based on hard data, as hard as any data in The Bell Curve. These groups repeatedly tested below average on the mental tests of the World War I era, both in the army and in civilian life. For Jews, it is clear that later tests showed radically different results--during an era when there was very little intermarriage to change the genetic makeup of American Jews.

Anyway! Your only responsibility is to do stuff that’s actually in Japanese; the remainder of the responsibility rests entirely with the Japanese stuff — media — itself. The media has a responsibility to entertain you. You don’t have to find the value in it; it has to demonstrate its value to you by being so much fun that you don’t notice time going by — by sucking you in. It has to make you wish that eating and sleep and bodily hygiene could take care of themselves because they cut into your media time. And if it doesn’t do that or it stops doing that, then you “fire” it by changing to something else. You are the boss and there are no labor laws. Fire the mother. You do the work of setting up and showing up to the environment, but after that the environment must work for you.

DO NOT, DO NOT, DO NOT turn Japanese into work. Don’t turn it into “study”; don’t turn it into 勉強 (a word that refers to scholastic study in Japanese, but actually carries the rather negative meaning of “coercion” in Chinese). Just play at it. PLAY. That’s why I keep telling people: don’t make all these rules about what is and is not OK for you to do in Japanese, or how Gokusen is over-coloured by the argot of juvenile delinquents or watching Love Hina will make you talk like a girl — it doesn’t matter, you need to learn all that vocabulary in order to truly be proficient in Japanese anyway, so whatever you watch is fine — as long as you’re enjoying it right now. Write this on your liver: just because anything is OK to watch in Japanese, that doesn’t mean that everything is worth watching…to you that is. One person’s Star Trek is another person’s…well, I can’t imagine how any human being could fail to love Star Trek, but you get the idea.

Most people think you build the product then you market it. Thinking in loops means you build the marketing into the product. The product doesn't precede the marketing. The product is the marketing.

The task of "making a thing satisfying our needs" as a single responsibility is split into two parts "stating the properties of a thing, by virtue of which it would satisfy our needs" and "making a thing guaranteed to have the stated properties". Business data processing systems are sufficiently complicated to require such a separation of concerns and the suggestion that in that part of the computing world "scientific thought is a non-applicable luxury" puts the cart before the horse: the mess they are in has been caused by too much unscientific thought.

Some time ago I visited the computing center of a large research laboratory where they were expecting new computing equipment of such a radically different architecture, that my colleagues had concluded that a new programming language was needed for it if the potential concurrency were to be exploited to any appreciable degree. But they got their language design never started because they felt that their product should be so much like FORTRAN that the casual user would hardly notice the difference "for otherwise our users won't accept it". They circumvented the problem of explaining to their user community how the new equipment could be used at best advantage by failing to discover what they should explain. It was a rather depressing visit.... The proper technique is clearly to postpone the concerns for general acceptance until you have reached a result of such a quality that it deserves acceptance. It is the significance of your message that should justify the care that you give to its presentation, it may be its "unusualness" that makes extra care necessary.

When someone asks if you have time for a meeting next Tuesday, you may have nothing on your calendar, so you say “sure.” If you hadn’t agreed to the meeting, you would have done something with that time - but what? By getting clear on what the “what” was that I could be doing made me better at saying no. When you say yes that one hour phone call next week, you are saying no to revamping your sales page or going to the gym or getting home an hour earlier. There is not always a right or wrong answer, but if you realize what you are saying no to every time you say yes, then you can make a judgement call: “Is this phone call more important to me than going to the gym today?”

We have come to a place where thanks to many libraries and frameworks, and overall improving software, what would’ve once used many developers to build from scratch is now more often than not, a bunch of people plumbing different things together. Software is creating software faster than we can use it. This is also why you are seeing so many of these “no-code” or “low-code” solutions pop up all over the place. There are increasingly fewer reasons to write code, and those who are writing code should, and do, increasingly write less of it. This will only be more accelerated by shifting to remote work due to how it’s going to change how we decide what code to write.

Anyone who’s spent a few months at a sizable tech company can tell you that a lot of software seems to exist primarily because companies have hired people to write and maintain them. In some ways, the software serves not the business, but the people who have written it, and then those who need to maintain it. This is stupid, but also very, very true.

In a world where most employees are remote, this can be harder to do. Not only employees could be in touch with each other less, and in less personal ways, they might not be even able to do so without having non-monitored places. There will always be ways to employees to sneak around monitoring and surveillance, but it’ll be harder when everything is fully remote, and you’ll have less trust in those who will bond (or conspire with, depending on your POV) with you.

The remote-first mentality will be a god-send simply because you’ll no longer be restricted to a tiny piece of land with a questionable housing policy to source your talent. People estimate 40% of all VC funding going to landlords in the Bay, and I think that’s too conservative.

Obviously, things can get quite weird when you take this model to its logical end. In the Bay Area, where the companies are giant, the geography tiny and the housing policies extremely questionable, this has resulted in salaries ballooning to insane levels. Getting a six-figure salary straight out of college barely raises an eyebrow anymore at many big firms. Companies have gone to great lengths, including some illegal ones to curb this competitive behavior to depress the salaries.

Most people would like to believe salaries are determined by a cost-plus model, where you get a tiny bit less than the value you add to the company. However, in reality, they are really determined by the competition. Companies are forced to pay as much as possible to keep the talent for leaving. In a competitive labor market, this is often a good thing for the employees.

With limited or no access to technol-ogy, limited capacity to cope and adapt, limited or no savings, inadequate access to social services, and un-certainty about their legal status and potential to ac-cess healthcare services, tens of thousands of migrants and non-nationals have left Thailand over the past weeks.

Insight through making suggests that you’ll need to make simultaneous progress in theory-space and system-space to spot the new implications in their conjoined space. Effective system design requires insights drawn from serious contexts of use: you must constantly instantiate new theoretical ideas in new systems, then observe their impact in some serious context of use.

Whether in music (Bach, Lennon), art (Picasso, Bernini), film (Tarantino, Anderson), games (Blow, Lantz), fiction (Kundera, Tolstoy), the most eminent work is usually the result of a single person’s creative efforts. Occasionally it’s a very small group (Eames, Wrights).

Per Michael: you probably would rather have Stradivarius make your violin than Joshua Bell, but you’d probably rather hear Joshua Bell play. Each activity—violin-making and violin-playing—requires virtuosic skill and a lifetime of practice. It’s very unlikely to find both abilities in the same person!

One huge advantage to scaling up is that you’ll get far more feedback for your Insight through making process. It’s true that Effective system design requires insights drawn from serious contexts of use, but it’s possible to create small-scale serious contexts of use which will allow you to answer many core questions about your system.

WhyGeneral infrastructure simply takes time to build. You have to carefully design interfaces, write documentation and tests, and make sure that your systems will handle load. All of that is rival with experimentation, and not just because it takes time to build: it also makes the system much more rigid.Once you have lots of users with lots of use cases, it’s more difficult to change anything or to pursue radical experiments. You’ve got to make sure you don’t break things for people or else carefully communicate and manage change.Those same varied users simply consume a great deal of time day-to-day: a fault which occurs for 1% of people will present no real problem in a small prototype, but it’ll be high-priority when you have 100k users.Once this playbook becomes the primary goal, your incentives change: your goal will naturally become making the graphs go up, rather than answering fundamental questions about your system.

This is obviously a powerful playbook, but it should be deployed with careful timing because it tends to freeze the conceptual architecture of the system.

Part of the problem of social media is that there is no equivalent to the scientific glassblowers’ sign, or the woodworker’s open door, or Dafna and Jesse’s sandwich boards. On the internet, if you stop speaking: you disappear. And, by corollary: on the internet, you only notice the people who are speaking nonstop.

If painting is an aesthetic medium of vision, music an aesthetic medium of sound, and cooking an aesthetic medium of taste, then games are an aesthetic medium of action, Frank Lantz observes.

Annotations—even inline marginalia which include your own writing—have very little informational value. They’re atomized; they don’t relate to each other; they don’t add up to anything; they’re ultra-compressed; they’re largely unedited. That’s fine: think of them as just a reminder. They say “hey, look at this passage,” with a few words of context to jog your memory about what the passage was about.Since you’re going to write lasting notes anyway, annotations need carry just enough information to recreate your mental context in that moment of reading. You wouldn’t want to rely on that long-term, since then you’d just have a huge pile of hooks you’d have to “follow” anytime you wanted to think about your experience with that book.

Update 2020-01-14: I now store my outlines as Structure Zettel. For more information what a Structure Zettel is see this post.

Instead of having a task like “write an outline of the first chapter,” you have a task like “find notes which seem relevant.” Each step feels doable. This is an executable strategy (see Executable strategy).

In a classroom or professional setting, an expert might perform some of these tasks for a learner (Metacognitive supports as cognitive scaffolding), but when a learner’s on their own, these metacognitive activities may be taxing or beyond reach.

To successfully learn something new, people must evaluate their understanding, monitor for confusion or inconsistency, plan what to do next based on those observations, and coordinate that plan’s execution. This often falls under the category of “metacognition,” though prefer to unbundle its phenomena.

Ericsson claims (2016, p. 98) that there is no deliberate practice possible for knowledge work because there are no objective criteria (so, poor feedback), because the skills aren’t clearly defined, and because techniques for focused skill improvement in these domains aren’t known.

One common choice is to set daily goals for a certain number of hours at work. Success with this strategy requires a clear theory of how those hours will inexorably accumulate to the desired outcome. Simply spending some number of hours on a project is a fairly weak constraint: it’s easy to work with focus many hours unproductively.

Incremental writing is a method of writing in which ideas are written down and assembled incrementally. Incremental writing requires no linearity. It adapts to your way of thinking. Many great writers and scientists of the past used a variant of incremental writing using their own systems of notes. In SuperMemo, incremental writing is integral with the creative process and learning itself

By contrast, when we’re working on a large work-in-progress manuscript, we’re juggling many ideas in various states of completion. Different parts of the document are at different levels of fidelity. The document is large enough that it’s easy to lose one’s place or to forget where other relevant points are when one returns. Starting and stopping work for the day feel like heavy tasks, drawing heavily on working memory.

Instead, nurture the wild idea and let it develop over time by incrementally writing Evergreen notes about small facets of the idea.

The issue of the different layers is similar. If you chose software that doesn’t deal with those layers in a sophisticated way, you will not reap the benefits in the long term. Your archive will note work as a whole. I think that this is one of the reasons why many retreat to project-centered solutions, curating one set of notes for each book, for example. The problems that come with big and organic (= dynamic and living) systems is avoided. But so is the opportunity to create something that is greater than you.

After a while, I did not only have structure notes that structure content notes, I also had structure notes that mainly structured sets of structure notes. They became my top level structure notes because they began to float on the top of my archive, so to say.

Structure notes share a similarity to tags: Both point to sets of notes. Structure notes just add another element. They are sets with added structure. This added structure provides a better overview and adds to the utility of the archive.

But after a while, you won’t be able to keep up. When I search for tags I get a couple hundred of notes. I have to review them to connect a note to some of them, or get a grasp of what I wrote and thought about a specific topic. Naturally, a need to organize the archive arises at this point. I can’t remember how many notes I had when I experienced this. I introduced hub-like notes when I had between 500 and 700 notes.1 I gave myself an overview of the most important notes on that topic.