Across the world, universities have become obsessed with their position in global rankings (such as the Shanghai Ranking and Times Higher Education's list), even when such lists are based on what are, in our view, inaccurate data and arbitrary indicators.

Lately, metrics related to social usage and online comment have gained momentum — F1000Prime was established in 2002, Mendeley in 2008, and Altmetric.com (supported by Macmillan Science and Education, which owns Nature Publishing Group) in 2011.

The problem is that evaluation is now led by the data rather than by judgement.

Assessing staff solely on the basis of quantitative metrics is never acceptable, no matter what type of metric is being used

The two metrics used were a five-year average of research income compared with that of researchers at similar universities, and a score called field-weighted citation impact, which measures how often research papers are cited relative to the rest of the papers in their field.

membership of external bodies

The university declined to specify what these metrics were.

The debate highlights broader unease about the use of metrics in science as more data are collected to assess the quality of researchers’ work. Some say these quantitative measures of performance concentrate too much on publication records while failing to acknowledge other types of work, including teaching, committee work and peer review.

broader unease about the use of metrics in science

researchers are already encouraging improved practices in research assessment

The outputs from scientific research are many and varied, including: research articles reporting new knowledge, data, reagents, and software; intellectual property; and highly trained young scientists.

What we find as we crawl out of the hole is the challenge: Will we find that there are things that we believe that aren’t true? That’s quite possible.

One of the responses to the last paper was, “Oh, my God, all you’ll do is end up flooding the literature with cheap useless replications” — well, that’s not a problem right now. If it was the case that every Ph.D. had done replications as part of training, yes, we’d have many hundreds of them. Fantastic. You could have a special open access journal called informatics replication, for goodness’ sake, to put the less major ones in. These are just non-problems.

We don’t have a culture that recognizes or rewards replication work, so there’s no good practice around what good replication looks like. Whereas I think in psychology, for example, you’ve seen great progress since the early concerns. But that took a lot of really public failures of studies.

It’s a near universal challenge. The question is whether those differences are a foundational barrier to science, or whether they’re manageable. And I think most people who look at this area would say, look, they’re manageable.

The first reaction that’s really strong is, “Well, you can’t do replication in digital health because it’s all so special.” To which I say, I’m sorry, this is not the case. It’s a good excuse, but it won’t hold. And also, if it was the case, my God, what a disaster it would be, because that would mean there was no science to what we’re doing.

the field has yet to build a culture that values scientific best practices

Over six months, he and colleague Huong Ly Tong dredged up all the journal-published papers they could find analyzing the outcomes of clinical decision support systems. They found 4,063 — and of those, only 12 were replications.

Most recently, work pointed to flaws in an algorithm to predict the risk of sepsis, integrated into Epic’s electronic health record platform. A recent STAT investigation found those shortcomings extend to other Epic algorithms, including those used by hospitals to predict how long patients will be hospitalized or who will miss appointments.

Despite the problems with the algorithms, Epic incentivizes hospitals to adopt them.

Like many other groups that build health algorithms, Epic doesn’t publicly share details of how the algorithms are built. Researchers at hospitals that use Epic are able to scrutinize the tools, but any investigations are challenging: they can’t disclose proprietary information

one could imagine a world in which failure to produce a machine-readable structured procedure report precluded being paid at all

If we had truly robust standards for electronic data interchange and less anxiety about privacy, these kinds of data could be moved around more freely in a structured format. Of course, there are regional exchanges where they do. The data could also be created in structured format to begin with.

Supply chains—starting with the factories upstream, running through the ports and rail yards and warehouses, and ending with retail—are large and complex systems. These systems need to be adaptive, and yet the news shows us they are not. 

Most decision-making systems have trouble with unexpected shifts in data. They are trained to make decisions in some contexts, and they break when something unforeseen happens. They are brittle.

If we say someone is smart, we rarely mean that they can recognize faces. We very often mean that they know what to do to reach their goal.

Most machine learning algorithms are good at perceptive tasks that would take a person under a second to perform, such as recognizing a voice or a face. But deep reinforcement learning can learn tactical sequences of actions, things like winning a board game or delivering a package.

the STARD guidelines were updated in 2015 and are a set of 30 essential items that should be part of any diagnostic accuracy study; from sample size calculations to cross tabulation of results. The meta-analysis found that only 24/273 studies mentioned adherence to guidelines (interestingly the authors don't say if they actually were adherent or not) or contained a STARD flow diagram.

Unfortunately the findings from the second point almost completely undermine the first, and so that's what I'll be focusing on.

A deep learning model was used to render a prediction 24 hours after a patient was admitted to the hospital. The timeline (top of figure) contains months of historical data and the most recent data is shown enlarged in the middle. The model "attended" to information highlighted in red that was in the patient's chart to "explain" its prediction. In this case-study, the model highlighted pieces of information that make sense clinically.

An “attention map” of each prediction shows the important data points considered by the models as they make that prediction.

We emphasize that the model is not diagnosing patients — it picks up signals about the patient, their treatments and notes written by their clinicians, so the model is more like a good listener than a master diagnostician.

Before we could even apply machine learning, we needed a consistent way to represent patient records, which we built on top of the open Fast Healthcare Interoperability Resources (FHIR) standard as described in an earlier blog post.

ill-structured problems

No matter an AI's final Turing test score, a script built to imitate human conversation or recognize patterns isn't something we'd ever describe as being truly intelligent. And that goes for other major AI milestones: IBM's Deep Blue is better at chess than any human and Watson proved it could outsmart Jeopardy world champions, but they don't have any consciousness of their own.

With Ex Machina, the directorial debut of 28 Days Later and Sunshine writer Alex Garland, we can finally put the Turing test to rest. You've likely heard of it -- developed by legendary computer scientist Alan Turing (recently featured in The Imitation Game), it's a test meant to prove artificial intelligence in machines. But, given just how easy it is to trick, as well as the existence of more rigorous alternatives for proving consciousness, passing a test developed in the '50s isn't much of a feat to AI researchers today.

As originally conceived, the Turing test involves a natural language conversation between a machine and human conducted through typed messages from separate rooms. A machine is deemed sentient if it manages to convince the human that it's also a person; that it can "think."

That's why we will always stay smarter than AI.

People will always be faster to adjust than computers, because that's what humans are optimized to do

For Booking.com, those new categories could be defined in advance, but a more general-purpose AI would have to be capable of defining its own categories. That's a goal Hofstadter has spent six decades working towards, and is still not even close.

perception is far more than the recognition of members of already-established categories — it involves the spontaneous manufacture of new categories at arbitrary levels of abstraction

This concept of context is one that is central to Hofstadter's lifetime of work to figure out AI

They may identify attributes such as 'ocean', 'nature', 'apartment', but Booking.com needs to know whether there's a sea view, is there a balcony and does it have a seating area, is there a bed in the room, what size is it, and so on. Dua and his colleagues have had to train the machines to work with a more detailed set of tags that matches their specific context.

A friend asked me whether Google Translate’s level of skill isn’t merely a function of the program’s database. He figured that if you multiplied the database by a factor of, say, a million or a billion, eventually it would be able to translate anything thrown at it, and essentially perfectly. I don’t think so. Having ever more 'big data' won’t bring you any closer to understanding, since understanding involves having ideas, and lack of ideas is the root of all the problems for machine translation today.

The bailingual engine isn’t reading anything — not in the normal human sense of the verb 'to read'. It’s processing text. The symbols it’s processing are disconnected from experiences in the world. It has no memories on which to draw, no imagery, no understanding, no meaning residing behind the words it so rapidly flings around.

Clearly Google Translate didn’t catch my meaning; it merely came out with a heap of bull. 'Il sortait simplement avec un tas de taureau.' 'He just went out with a pile of bulls.' 'Il vient de sortir avec un tas de taureaux.' Please pardon my French — or rather, Google Translate’s pseudo-French.

Humans are optimized for learning unlimited patterns, and then selecting the patterns we need to apply to deal with whatever situation we find ourselves in

Computers are much better than us at only one thing — at matching known patterns.

underestimate our own performance because we rarely stop to think how much we already know

apparent successes

Machine intelligence is still pretty dumb, most of the time. It's far too early for the human race to throw in the towel.

apparently

Not for the first time in its history, artificial intelligence is rising on a tide of hype.

The researchers started with 140,000 hours of YouTube videos of people talking in diverse situations. Then, they designed a program that created clips a few seconds long with the mouth movement for each phoneme, or word sound, annotated. The program filtered out non-English speech, nonspeaking faces, low-quality video, and video that wasn’t shot straight ahead. Then, they cropped the videos around the mouth. That yielded nearly 4000 hours of footage, including more than 127,000 English words.

They fed their system thousands of hours of videos along with transcripts, and had the computer solve the task for itself.

Easy reading only makes you informed; hard reading makes you competent. Easy reading prevents you from being ignorant; hard reading makes you smarter.

reading easy texts exclusively is prone to confirmation bias and exacerbates our blind spots. After all, if you’re ignorant (partially or entirely) of an opposing view, surely you wouldn’t think you’re “objective”?

Pocket is also free, but I never really got into using it. (Doesn’t fit my workflow)

Because that keeps us paying attention. That’s right, we’re still paying for “free” information.

The third rule is you should always make what you read your own.

to solve information overload — or more appropriately, attention overload — we need to create a reading workflow

we should do three things: Manage what we pay attention toManage how we pay attentionProcess them deeply

Ultimately, my dream—similar to that of Bush’s—is for individual commonplace books to be able to communicate not only with their users in the Luhmann-esqe sense, but also communicate with each other.

IndieWeb friendly building blocks like Webmention, feeds (RSS, JSON Feed, h-feed), Micropub, and Microsub integrations may come the closest to this ideal.

The idea of planting a knowledge “seed” (a note), tending it gradually over time with regular watering and feeding in a progression of Seedlings → Budding → Evergreen is a common feature.

Mike Caulfield’s essays including The Garden and the Stream: A Technopastoral

Second brain is a marketing term

one might consider some of the ephemeral social media stream platforms like Twitter to be a digital version of a waste book

They have generally been physical books written by hand that contain notes which are categorized by headings (or in a modern context categories or tags. Often they’re created with an index to help their creators find and organize their notes.

Some of the examples you describe – the extraordinary variance seen in sentencing for the same crimes (even influenced by such external matters as the weather, or the weekend football results), say, or the massive discrepancies in insurance underwriting or medical diagnosis or job interviews based on the same baseline information – are shocking. The driver of that noise often seems to lie with the protected status of the “experts” doing the choosing. No judge, I imagine, wants to acknowledge that an algorithm would be fairer at delivering justice?The judicial system, I think, is special in a way, because it’s some “wise” person who is deciding. You have a lot of noise in medicine, but in medicine, there is an objective criterion of truth.

If we look at the arc of the 20th century, heavier than air flight transformed our world in major ways.

Don’t be afraid to take courses first.

the goal is not to get every fact and detail inside your head, but to have a good map of the area so you know where to look when you need to find it again

Read everything you can, including making highlights of sections you think you may need to revisit later. If I finish a book or longer paper, I’ll often make a new document where I’ll pull notes and quotes from my original reading, as well as do my best to summarize what I read from memory. The goal here is partly to practice retrieval and understanding, but also partly to give yourself breadcrumbs so you can find things more easily later.

I don’t find those methods very helpful, but it’s possible that I’m simply inept at them.

it’s better to follow a breadth-first rather than depth-first search, since you can easily spend too much time going down one rabbit hole and miss alternate perspectives

When following citations, I look for two factors: frequency and relevance. Works that are cited frequently are more central to a field.

After reading about two dozen Kindle previews for the most relevant seeming ones

Literature Review, Meta-Analysis and Textbooks

Wikipedia is usually a good starting point, because it tends to bridge the ordinary language way of talking about phenomena and expert concepts and hypotheses. Type your idea into Wikipedia in plain English, and then note the words and concepts used by experts

Open-ended activity often languishes from a lack of completeness

Setting Up Scope and Topic

If your predictions don’t fit the actual data, you get a high prediction error that updates your internal model—to reduce further discrepancies between expectation and evidence, between model and reality. Your brain hates unfulfilled expectations, so it structures its model of the world and motivates action in such a way that more of its predictions come truer.

Your brain runs an internal model of the causal order of world that continually creates predictions about what you expect to perceive. These predictions are then matched with what you actually perceive, and the divergence between predicted sensory data and actual sensory data yields a prediction error.

If your brain is Bayesian, however, it doesn’t process sensory data like that. Instead, it uses predictive processing (also known as predictive coding)2 to predict what your eyes will see before you get the actual data from the retina.

Your brain is a prediction machine.

There are obvious benefits to AI systems that are able to automatically learn better ways of representing data and, in so doing, develop models that correspond to humans’ values. When humans can’t determine how to map, and subsequently model, values, AI systems could identify patterns and create appropriate models by themselves. However, the opposite could also happen — an AI agent could construct something that seems like an accurate model of human associations and values but is, in reality, dangerously misaligned.

Sometimes humans even value things that may, in some respects, cause harm. Consider an adult who values privacy but whose doctor or therapist may need access to intimate and deeply personal information — information that may be lifesaving. Should the AI agent reveal the private information or not?

A thermostat, for example, is a type of reflex agent. It knows when to start heating a house because of a set, predetermined temperature — the thermostat turns the heating system on when it falls below a certain temperature and turns it off when it goes above a certain temperature. Goal-based agents, on the other hand, make decisions based on achieving specific goals. For example, an agent whose goal is to buy everything on a shopping list will continue its search until it has found every item. Utility-based agents are a step above goal-based agents. They can deal with tradeoffs like the following: Getting milk is more important than getting new shoes today. However, I’m closer to the shoe store than the grocery store, and both stores are about to close. I’m more likely to get the shoes in time than the milk.” At each decision point, goal-based agents are presented with a number of options that they must choose from. Every option is associated with a specific “utility” or reward. To reach their goal, the agents follow the decision path that will maximize the total rewards.

As data-driven learning systems continue to advance, it would be easy enough to define “success” according to technical improvements, such as increasing the amount of data algorithms can synthesize and, thereby, improving the efficacy of their pattern identifications. However, for ML systems to truly be successful, they need to understand human values. More to the point, they need to be able to weigh our competing desires and demands, understand what outcomes we value most, and act accordingly.

objective function that tries to describe your ethics

The problem is, algorithms were never designed to handle such tough choices. They are built to pursue a single mathematical goal, such as maximizing the number of soldiers’ lives saved or minimizing the number of civilian deaths. When you start dealing with multiple, often competing, objectives or try to account for intangibles like “freedom” and “well-being,” a satisfactory mathematical solution doesn’t always exist.

many other systems that are already here or not far off will have to make all sorts of real ethical trade-offs

focusing on more conventional issues, since they’ll be what you’re most likely to come across. But these are unlikely to be your highest-impact options

Another much-debated question has been, ‘How does the agent’s choice of macro-block placements survive subsequent steps in the chip-design process?’ As mentioned earlier, human engineers must iteratively adjust their floorplans as the logic-circuit design evolves. The trained agent’s macro-block placements somehow evade such landmines in the design process, achieving superhuman outcomes for timing (ensuring that signals produced in the chip arrive at their destinations on time) and for the feasibility and efficiency with which wiring can be routed between components.

Mirhoseini et al. estimate that the number of possible configurations (the state space) of macro blocks in the floorplanning problems solved in their study is about 102,500. By comparison, the state space of the black and white stones used in the board game Go is just 10360.

Modern chips are a miracle of technology and economics, with billions of transistors laid out and interconnected on a piece of silicon the size of a fingernail. Each chip can contain tens of millions of logic gates, called standard cells, along with thousands of memory blocks, known as macro blocks, or macros. The cells and macro blocks are interconnected by tens of kilometres of wiring to achieve the designed functionality.

Persistent Identifiers (PIDs) beyond the DOI

To get closer to attaining coveted “rich” metadata and ultimately contribute to a “richer” scholarly communication ecosystem, journals first need to have machine-readable metadata that is clean, consistent, and as interoperable as possible.

To see this in action, visit somebody’s Open Ledger and recognize them for a scholarly contribution. For example, you could visit my Open Ledger: https://rescognito.com/0000-0002-9217-0407 and click the “Recognize” button at the top of the page to recognize me for a scholarly activity such as “Positive public impact”.

We also make it possible to generate and store assertions about activities that have no proximate physical or digital corollary, such as “mentoring” and “committee work”

No data transformations, no XML manipulation, no data synchronization, no DTD updates, no transfer between vendors, no coordination, no training, no management time! In addition, the assertion outputs are superior in granularity, provenance, specificity, display, and usability.

Another benefit of structured assertions is that they can be accessed via APIs

Because contributors are verifiably and unambiguously identified by their ORCID iD

The entire process has to be coordinated, managed, and synchronized — often over multiple continents

have the assertions be made by trusted, branded gatekeepers to guarantee their provenance

Think of it this way: the value is not in the content, it is in the assertions — scholarly publishers don’t publish content, they publish assertions. This provides a coherent model to explain why publishers add value even though the source content and peer review are provided for free. Journals incur legitimate costs generating and curating important and valuable assertions.

The authors have these conflicts of interest

The findings are supported by the data”, “The work is novel”, “The work builds-on (i.e., cites) other work”

The document was peer reviewed by anonymous reviewers

It was not plagiarized

The statistics are sound

Who funded the work

When was it released

Where was the work done

Who created the document

Publisher costs usually include copyediting/formatting and organizing peer review. While these content transformations are fundamental and beneficial, they alone cannot justify the typical APC (Article Publication Charge), especially since peer reviewers are not paid.

Journals like Science and Nature are financially viable and they create a kind of club. However, this is not a knowledge community in any meaningful sense. The authors of an article on the genome of an organism are not producing knowledge in concert with those of an article on the formation of stars. In these cases the “good” being produced is prestige, or brand value. Rather than being knowledge clubs, they are closer to “social network markets”, in which the choices that individuals make, such as where to seek to publish, are driven by the actions of those with higher prestige in the network. Such markets are effective means for extracting resources out of communities.

we propose that the value of a well-run journal does not lie simply in providing publication technologies, but in the user community itself. Journals should be seen as a technology of social production and not as a communication technology.

social life of journals and the knowledge communities they sustain

You’ll be unlikely to stick with and excel in any path in the long term if you don’t enjoy it and it doesn’t fit with the rest of your life.

the main components of a satisfying job are:A sense of meaning or helping othersA sense of achievementEngaging work with autonomySupportive colleaguesSufficient ‘basic conditions’ such as fair pay and non-crazy working hours

your long-term goal is to maximise the product of these three factors over the remainder of your career

we mean increasing wellbeing in the long term (the ‘positive impact’ part), and treating everyone’s interests as equal, no matter their gender, background, where they live, or even when they live (that’s the ‘impartial’ part).

it’s hard to agree on what the terms ‘help people’, ‘positive impact’, and ‘personally fulfilling’ actually mean — let alone craft goals that will help you achieve these things

We think that very often, much of what matters about people’s actions is the difference they make over the long term — basically because we think that the welfare of those who live in the future matters; indeed, we think it matters no less than our own.

What do we mean by “considered impartially”? In short, we mean that we strive to treat equal effects on different beings’ welfare as equally morally important, no matter who they are — including people who live far away or in the future, and including non-humans.

most of the ideas for what welfare might consist in — for instance happiness or getting to live a life of your choosing– are tightly correlated with one another

we think it makes sense to use the following working definition of social impact: “Social impact” or “making a difference” is about promoting welfare, considered impartially, over the long term — without sacrificing anything that might be of comparable moral importance.

When people value their attention and energy, they become valuable

the advertising-driven ones

Our world is shaped by humans who make decisions, and technology companies are no different…. So the assertion that technology companies can’t possibly be shaped or restrained with the public’s interest in mind is to argue that they are fundamentally different from any other industry

Career decision making involves so much uncertainty that it’s easy to feel paralysed. Instead, make some hypotheses about which option is best, then identify key uncertainties: what information would most change your best guess?

One of the most useful steps is often to simply apply to lots of interesting jobs.

by looking at how others made it

think of your career as a series of experiments designed to help you learn about yourself and test out potentially great longer-term paths

The returns of aiming high are usually bigger than the costs of switching to something else if it doesn’t work out, so it’s worth thinking broadly and ambitiously

ask what the world needs most

Being opportunistic can be useful, but having a big positive impact often requires doing something unusual and on developing strong skills, which can take 10+ years.

career capital

what helps the most people live better lives in the long term, treating everyone’s interests as equal

We’d encourage you to make your own definition of each.

Why educational technologies haven't transformed the college experience

ABSTRACT

promises of digital “dataism” are thwarted by the entrenched temporal organisation of schooling, and teacher-centred understandings of students as coerced subjects

Using sociological theorisation of institutional logics

Perhaps for everyone, a moment or occasion of leadership will emerge, reveal itself, and call to us with the painful, necessary task of speaking up, patiently asking for alternatives, insistently rocking the boat

Ivan Illich, no fan of schooling or authoritarian structures of any kind, writes movingly about the role of the true, deep teacher. So does George Steiner, using language of “master” and “disciple” that would make many open-web folks cringe–or worse. Yet even the great and greatly democratic poet Walt Whitman salutes his “eleves” at one point. And I have experienced and been very grateful for the wisdom of those teacher-leaders who brought me into a fuller experience and understanding of my own responsibilities as a leader.

leading is risky business

but is about writerly choices

Getting to grips with structure means keeping your reader in mind.

it’s more accurate to say that readers notice the absence of structures, and/or when we shift the logics of one structure to another mid-stream, without saying anything.

The more I used Roam, the more valuable my notes became. This increase in value made me realize that I needed a little more trust in the developers' approach to release management. It wasn't always clear how changes would affect my workflow.

I also related old notes on similar topics to the Kanban concepts. In some cases, I saw a detail from Getting Things Done in a new light and took note about that

What was your reading intent and how can you capture it best?

Judgments made by different people are even more likely to diverge. Research has confirmed that in many tasks, experts’ decisions are highly variable: valuing stocks, appraising real estate, sentencing criminals, evaluating job performance, auditing financial statements, and more. The unavoidable conclusion is that professionals often make decisions that deviate significantly from those of their peers, from their own prior decisions, and from rules that they themselves claim to follow.

rationale for the decision taken

particular benefit to authors for whom English is not a first language

an AI tool which screens papers prior to peer review could be used to advise authors to rework their paper before it is sent on for peer review.

more inclined to reject papers based on this negative first impression derived from what are arguably relatively superficial problems.

One possible explanation for the success of this rather simplistic model is that if a paper is presented and reads badly, it is likely to be of lower quality in other, more substantial, ways, making these more superficial features proxy useful metrics for quality.

most journals still insist on submissions in .docx format.

The most common way to stage an argument in the thesis goes something like this: Here is a puzzle/problem/question worth asking. If we know more about this puzzle/problem/question then something significant (policy, practice, more research) can happen.Here is what we already know about the puzzle/problem/question. I’ve used this existing knowledge (literatures) to help: my thinking and approach; my research design; make sense of my results; and establish where my scholarly contribution will be. Here is how I designed and did the research in order to come up with an “answer”.Here’s the one/two/three clusters of results.Missing stepNow here’s my (summarised) “answer” to the puzzle/problem/question I posed at the start. On the back of this answer, here’s what I claim as my contribution(s) to the field. Yes I didn’t do everything, but I did do something important. Because we now know my answer, and we didn’t before I did the research, then here are some possible actions that might arise in policy/practice/research/scholarship.

Imagine you could take those Lego pieces and put them in any place you wanted: in emails, in chats, in other apps,

we must shed our outdated concept of a document. We need to think in terms of flexible jumping and viewing options. The objects assembled into a document should be dealt with explicitly as representaions of kernel concepts in the authors' minds, and explicit structuring options have to be utilized to provide a much enhanced mapping of the source concept structures.

It really gets hard when you start believing in your dreams.

Draft notes, E-mail, plans, source code, to-do lists, what have you

editor-browser tool sets

Many years ago, I dreamed that digital technology could greatly augment our collective human capabilities for dealing with complex, urgent problems. Computers, high-speed communications, displays, interfaces — it's as if suddenly, in an evolutionary sense, we're getting a super new nervous system to upgrade our collective social organisms. I dreamed that people were talking seriously about the potential of harnessing that technological and social nervous system to improve the collective IQ of our various organizations.

we miss the deep understanding that comes from dialogue and exploration.

The medium should allow people to think with their bodies, because we are more than fingers and hands.

“With no one telling me what to work on, I had to decide for myself what was meaningful in this life. Because of how seriously I took my work, this process was very difficult for me,”

most professional programmers today spend their days editing text files inside an 80-column-wide command line interface first designed in the mid-1960s.

Commercial apps force us into ways of working with media that are tightly prescribed by a handful of people who design them

“Every Representation of Everything (in progress),” showcasing examples of different musical notation systems, sign languages, mathematical representations, chemistry notations.

Bret Victor, the engineer-designer who runs the lab, loves these information-rich posters because they break us out of the tyranny of our glassy rectangular screens.

“When we had electromechanical systems, we used to be able to test them exhaustively,” says Nancy Leveson, a professor of aeronautics and astronautics at the Massachusetts Institute of Technology who has been studying software safety for 35 years. She became known for her report on the Therac-25, a radiation-therapy machine that killed six patients because of a software error. “We used to be able to think through all the things it could do, all the states it could get into.” The electromechanical interlockings that controlled train movements at railroad crossings, for instance, only had so many configurations; a few sheets of paper could describe the whole system, and you could run physical trains against each configuration to see how it would behave. Once you’d built and tested it, you knew exactly what you were dealing with.

Leibniz’s notation, by making it easier to do calculus, expanded the space of what it was possible to think.

As science becomes more about computation, the skills required to be a good scientist become increasingly attractive in industry. Universities lose their best people to start-ups, to Google and Microsoft. “I have seen many talented colleagues leave academia in frustration over the last decade,” he wrote, “and I can’t think of a single one who wasn’t happier years later.”

maybe computational notebooks will only take root if they’re backed by a single super-language, or by a company with deep pockets and a vested interest in making them work. But it seems just as likely that the opposite is true. A federated effort, while more chaotic, might also be more robust—and the only way to win the trust of the scientific community.

“Frankly, when you do something that is a nice clean Wolfram-language thing in a notebook, there’s no bullshit there. It is what it is, it does what it does. You don’t get to fudge your data,” Wolfram says.

“The place where it really gets exciting,” he says, “is where you have the same transition that happened in the 1600s when people started to be able to read math notation. It becomes a form of communication which has the incredibly important extra piece that you can actually run it, too.”

What I’m studying is something dynamic. So the representation should be dynamic.”

Victor has long been convinced that scientists haven’t yet taken full advantage of the computer. “It’s not that different than looking at the printing press, and the evolution of the book,” he said. After Gutenberg, the printing press was mostly used to mimic the calligraphy in bibles. It took nearly 100 years of technical and conceptual improvements to invent the modern book. “There was this entire period where they had the new technology of printing, but they were just using it to emulate the old media.”

Papers may be posted online, but they’re still text and pictures on a page.

There was no public forum for incremental advances.

The pandemic has forced everyone to become video editors and generally not very good ones.

providing context rather than content

focus on creating communities

Teachers feel bound by tradition to deliver content and the students expect the teacher to deliver content and it's very hard to escape from this mindset.

My assertion is based on the observation that a great deal of learning does take place in connective environments on the world wide web, that these have scaled to large numbers, and that often they do not require any institutional or instructional support.

community

Offense, insult, and hurt feelings are not particularly important

life contains more suffering than happiness

Some things get better and some things get worse.

generally up to you