#3 The foundations of humane technology: Minimizing harmful consequences

I’m taking at my pace the free online course “Foundations of humane technology” by the Center for Humane Technology (whom I had heard of after watching the excellent documentary “The Social Dilemma”.) Here are my notes.


“Our economic activity is causing the death of the living planet and economists say, ‘Yeah, yeah, that’s an environmental externality.’ There’s something profoundly wrong with our theories if we’re dismissing or just happy to label the death of the living planet as an externality.”

Kate Raworth, “renegade economist”

Negative externalities are unaccounted side effects that result in damages or harms. At small scale they may be acceptable, but aggregated they may be catastrophic. They are external in that the company causing them does not pay the costs. Social, health-related, and environmental costs usually end up borne by society, now or in the future.

Example 1: We can regularly upgrade to the latest exciting smartphone. Externality: 50 million tons of toxic e-waste globally per year (source).

Example 2: The idiots who live in the house behind mine, including their large dog, regularly make a lot of noise, day or night, despite my many complaints over the years. Externality: additional cost of my cranking up the ventilator so the resulting white noise may cover theirs.

🤔 Personal reflection: Externalities

What are the conversations required to identify externalities in your own work? In what ways has your company, organization, or industry already learned lessons about externalities?

There is a proven feedback loop in place as part of our work process (https://www.w3.org/Consortium/Process/) which ensures that any issue is surfaced during the stages of development of open web standards. Externalities then may become work items themselves, or may be specific to particular specifications only. Conversations take place in the open in our multi-stakeholders forum.

Our organization was founded as an industry consortium a few years after the Web took off, by the inventor of the Web himself, to collaboratively and globally create the protocols and guidelines needed for the web to be a universal and agnostic platform which would guarantee its interoperability and availability for everyone. The way the consortium evolved over the years is a testament about learned lessons about externalities.
For example, in 1997, 2.5 years into its existence, W3C launched the International Web Accessibility Initiative (https://www.w3.org/Press/WAI-Launch.html), to remove accessibility barriers for all people with disabilities, with the endorsement of then USA President Bill Clinton who wrote “Congratulations on the launch of the Web Accessibility Initiative — an effort to ensure that people with disabilities have access to the Internet’s World Wide Web.” (https://www.w3.org/Press/Clinton.html)
25 years after, the W3C’s WAI is still very active as Web accessibility goals evolve just as Web technologies are created.

🤔 Personal reflection: Externalities at scale

In an effort to help users celebrate the people in their lives, a photo sharing app unveils a “friend score” that correlates with the people whose posts you most engage with on the app. How might this situation generate serious externalities when scaled to millions or billions of users? For example, how might they impact personal relationships, mental health, shared truth, or individual well-being?

Externalities: I can think of two exernalities: 1) What I post is influenced in a way that may either be reinforcing the feedback loop or breaking it, but in either cases I no longer retain control because what I post becomes driven by the score. 2) This is likely to antagonize myself with the subset of friends which do not feature or feature insufficiently in the score.

In our economic systems

“unchecked economic growth can destroy more value than it creates”

Center for Humane Technology

A few concepts:

  • Extraction occurs when a company removes resources from an environment more quickly than they are replenished.
  • On the other hand, stewardship is about creating value by helping value thrive in place.
  • To create long-term value, we must balance efficiency and scale with resilience, the ability of a complex system to adapt to unexpected changes. When we steward the complex systems that we rely on, they return the favor by supporting us in unexpected ways when crisis hits.

Over-extracting may lead to collapse

Our economic systems tend to operate at an unsustainably extractive scale:

  • prioritize growth at all costs, at the expense of environmental damage, or exploitation of labor forces;
  • are based on notions such as “Nature is a stock of resources to be converted for human purposes“;
  • foster competitive behaviours where even companies that understand the harms of over-extraction and wish to chart a different path face a harsh reality: to stay competitive everyone keeps being engaged, in eve in harmful behavior not because they want to, but because if they don’t, someone else in the market will.

“Often the best way to escape a [competitive behaviour] trap is to band together and change the rules. Competition must be balanced with collaboration and regulation, especially in places where extraction at scale creates widespread risks.

Just as businesses need markets to compete in, they need movements to collaborate in, especially when those businesses are values-driven.”

Center for Humane Technology

🤔 Personal reflection: Assess externalities

Imagine: Your product, which helps creators, mostly teens, express themselves to their friends and broader public audiences, is extremely good at training people to create compelling content and rewards them socially for doing so, but has contributed to a new externality: Reports about severe mental health struggles among influencers (maintaining a large, engaged audience is causing burn out, anxiety, social isolation, etc.) Yet young people are hooked on the idea: “social media influencer” is now the fourth-highest career aspiration among elementary school students.

  1. Scale of the externality. How widespread will it be?
    One out of four is already quite widespread. Compounded with the fact that teens tend to obsess unreasonably over fads, this could spread even more to toxic levels.
  2. Stakes of the impact. For example, is the impact irreversible?
    There is a risk of real harm, including self-harm, given the small number of potentially successful social media influencers, or the relatively short span of the success window. The impact may include lost opportunities to pursue a path of more sustainable livelihood.
  3. Vulnerability of the group or system impacted. How exposed is it?
    Teens are very vulnerable because easily influenced. They are at a critical point in their development on the track to adulthood where the choices they make are likely to shape and define them on the long-term.
  4. Long-term costs. If this externality is left unaddressed, who will bear the cost and how costly will it be?
    Like over-extraction, the value will diminish as the market is flooded. The race to the social media influence likely clears a path with less competition for other careers, but for as long as it takes for the race to lose its appeal, many will be left losing while very few succeed.
  5. Paths to reduction. How might less of this externality be created?
    Designing more around creating compelling content and less around social rewarding.

badge earned: "minimize harmful consequences"

The ultimate goal should not be to have companies pay to mitigate the harms their products create—it should be to avoid creating harm in the first place.

Center for Humane Technology

People and safety over profits, please

When algorithms prioritize content based on engagement, the most harmful and engaging content goes viral.

In the case of social media, it takes many orders of magnitude more effort to verify a fact than to invent a lie. Even a company spending billions on fact checking will always have their team outnumbered by those creating disinformation. For this reason, creating structural changes to disincentivize disinformation will almost always be more effective than hiring fact checkers to address the problem, even though fact-checking is an important part of the solution.

“Facebook’s Integrity Team researchers found that removing the reshare button after two levels of sharing is more effective than the billions of dollars spent trying to find and remove harmful content.”

Centre for Humane Technology, #OneClickSafer campaign

The Yak-layering problem

If yak-shaving is the masterful art of removing problems one by one until you eventually get to what you originally wanted to fix, then yak-layering is the unfortunate piling on top of each other of unintended consequences, which may become after years quite complex, obscured and often difficult to change.

"What are you working on?"
[person at computer answers:] "Trying to fix the problems I created when I tried to fix the problems I created when I tried to fix the problems I created when..."

Less is more

Addressing harmful externalities by doing less of the activities that generate them gives humans and our ecosystem a chance to get healthy again. So, while it’s easy to think that more technology can solve our problems, rather than creating a technological solution to address an externality, we can work to reduce the externality itself.

For example, implementing energy efficiency programs or appliances so that we use less energy is often cheaper and more environmentally friendly than generating more energy from renewable sources.

Paradigm shifting

current paradigmparadigm shifting
negative externalitiesturn into design criteria
profit/growth at all costsbind scale to responsibility
fix tech with more techcreate fewer risks
designdesign for the better
hiding/ignoring externalitiesadd mitigations to road-map
traditional success metrics (KPIs)align with your values

Day-to-day work at W3C

An author for IEEE asked me last week, for an article he’s writing, for a high level introduction to the World Wide Web Consortium, and what its day-to-day work looks like.

Most of the time when we get asked, we pull from boilerplate descriptions, and/or from the website, and send a copy-paste and links. It takes less than a minute. But every now and then, I write something from scratch. It brings me right back to why I am in awe of what the web community does at the Consortium, and why I am so proud and grateful to be a small part of it.

Then that particular write-up becomes my favourite until the next time I’m in the mood to write another version. Here’s my current best high level introduction to the World Wide Web Consortium, and what its day-to-day work looks like, which I have adorned with home-made illustrations I showed during a conference talk a few years ago.

World Wide What Consortium?

The World Wide Web Consortium was created in 1994 by Tim Berners-Lee, a few years after he had invented the Word Wide Web. He did so in order for the interests of the Web to be in the hands of the community.

“If I had turned the Web into a product, it would have been in people’s interest to create an incompatible version of it .”

Tim Berners-Lee, inventor of the Web

So for almost 28 years, W3C has been developing standards and guidelines to help everyone build a web that is based on crucial and inclusive values: accessibility, internationalization, privacy and security, and the principle of interoperability. Pretty neat, huh? Pretty broad too!

From the start W3C has been an international community where member organizations, a full-time staff, and the public work together in the open.

W3C is the international standardisation body for Web Standards that operates for one Web, for all, everywhere.
Its motto is "leading the Web to its full potential".
In order to accomplish this, the W3C leverages its Process Document and Patent Policy.
W3C proceeds according to principles of open standardisation based on consensus and transparence.
Working Groups are at the heart of the W3C. There are currently 52, plus the W3C Advisory Board and the Technical Artchitecture Group, whose participants are elected W3C Members.
Each group gathers participants delegated by W3C Members, plus invited experts from the public. They contribute respectively to tests and implementations, reviews, comments and translations, and grant royalty-free. rights on technology they develop. Today there are 400 standards upon which the web relies. 
The W3C team is made of a team of 56.
W3C overview

The sausage

In the web standards folklore, the product –web standards– are called “the sausage” with tongue in cheek. (That’s one of the reasons behind having made black aprons with a white embroidered W3C icon on the front, as a gift to our Members and group participants when a big meeting took place in Lyon, the capital of French cuisine.)

Since 1994, W3C Members have produced 454 standards. The most well-known are HTML and CSS because they have been so core to the web for so long, but in recent years, in particular since the Covid-19 pandemic, we’ve heard a lot about WebRTC which turns terminals and devices into communication tools by enabling real-time audio/video, and other well-known standards include XML which powers vast data systems on the web, or WebAuthn which significantly improves security while interacting with sites, or Web Content Authoring Guidelines which puts web accessibility to the fore and is critical to make the web available to people of all disabilities and all abilities.

The sausage factory

The day to day work we do is really of setting the stages to bring various groups together in parallel to progress on nearly 400 specifications (at the moment), developed in over 50 different groups.

There are 2,000 participants from W3C Members in those groups, and over 13,000 participants in the public groups that anyone can create and join and where typically specifications are socialized and incubated.

There are about 50 persons in the W3C staff, a fourth of which dedicate time as helpers to advise on the work, technologies, and to ensure easy “travel” on the Recommendation track, for groups which advance the web specifications following the W3C process (the steps through which specs must progress.)

The illustration contains a stick figure with 15 busy arms and smaller images describing situations such as coding, negotiation, mastering the Process Document, the Recommendation Track and the Patent Policy.
Characteristics of the W3C Staff Contact:
* "super interface"
* represents the Director and the staff in their groups
* participant and contributor
* technical expert
* masters the process
* creates groups
* manages groups
* inter-group technical liaison
* consensual
Role of the W3C staff in work groups

The rest of the staff operate at the level of strategy setting and tracking for technical work, soundness of technical integrity of the global work, meeting the particular needs of industries which rely on the web or leverage it, integrity of the work with regard to the values that drive us: accessibility, internationalization, privacy and security; and finally, recruiting members, doing marketing and communications (that’s where I fit!), running events for the work groups to meet, and general administrative support.

With its Director Tim Berners-Lee, inventor of the Web; and its CEO Jeff Jaffe, the W3C team is made of over 50 people.
The W3C team is almost equally divided between technical and support, including administration, communications, business development, systems team, legal, Member satisfaction, global participation and community management.
The technical part is now divided between four functions:
Strategy to determine the priorities of the Consortium and new work.
Architecture and technology to ensure cohesion of architecture as well as technical choices.
Industry to define the vision of the Consortium according to the need of the Industry, which is shaped by the Web.
Projects to ensure timelines are met in delivering Web standards and support the working groups being successful.
W3C team

Why does it work?

Several of the unique strengths of W3C are our proven process which is optimized to seek consensus and aim for quality work; and our ground-breaking Patent Policy whose royalty-free commitments boosts broad adoption: W3C standards may be used by any corporation, anyone, at no cost: if they were not free, developers would ignore them.

From an idea to a standard
On the one hand, the public; on the other, W3C Members.
Each contribute differently to the standardisation process.
Standards progress this way:
A standard may originate from a W3C Workshop or incubation in a W3C Community Group. Both are open to the public and W3C Members.
Based on consensus, a W3C Working Group can be created.
A standard may originate from a W3C Member Submission, without requiring incubation or discussion at a Workshop.
When a Working Group is created, W3C Members delegate one or more participants and agree to make available under W3C royalty-free licensing commitments all of the work they do in that particular Working Group.
Individuals from the public participate as Invited Experts.
During the standardisation process, Members of the W3C contribue reviews, tests and implementations. The Public contributes comments during phases of public review.
After 3 to 6 years on average, a Web standard is born when a technology reaches the status of Recommendations. Everyone parties like mad [drawing of confettis] W3C Members contribute to press releases and perform promotion.
The public and the Members help with the translation effort as well as maintenance via errata.
From an idea to a standard

There are other strengths but in the interest of time, I’ll stop at the top two. There are countless stories and many other facets, but that would be for another time.

Sorry, it turned out to be a bit long because it’s hard to do a quick intro; there is so much work. If you’re still with me (hi!), did you learn anything from this post?

Book: “Persuasion” by Jane Austen

Ah, the wit of Jane Austen is sharp in this novel, and enjoyable as always. 

It is laid out a lot like a theatrical play. Many of the scenes could be played in their own stage. Except perhaps the long walks and the beach strolls. 

This novel is her shortest, I think. However, I was stricken by the over-abundance of the word “and”.

I may return to this post with more to add. I have only finished it now after starting it yesterday, and I may need to let it sink.

2022-07-27 update: “and” appears 2802 times in the span of 227 pages, and 24 chapters (that’s 12.3 per page) (*)

(*) [After starting to underline them in my book, I found it tedious and unreliable, so I found an HTML version of the book, stripped it of non-novel cruft using emacs and then piped a word count to a grep, embracing the nerddom, but then ran a better grep(**) command which Bert supplied and explained, because the simpler one would find hand, grand or wander, but not And,]

 (Wed, 27 Jul 2022 01:11:29 CET)-(koalie@gillie:~:)$grep and /Users/koalie/Library/Mobile\ Documents/com\~apple\~CloudDocs/Downloads/Persuasion\,\ by\ Jane\ Austen.html | wc -l

(**) (Wed, 27 Jul 2022 07:11:31 CET)-(koalie@gillie:~:)$grep -E -i -o '\band\b' /Users/koalie/Library/Mobile\ Documents/com\~apple\~CloudDocs/Downloads/Persuasion\,\ by\ Jane\ Austen.html | wc -l

(where -E = enable regexps, -i = case-insensitive, -o = put every occurrence on a separate line, \b = word edge) [Thanks Bert!]