Notes from 1/9–8/9

This new research clearly demonstrates that bodily response is not the only means by which acupuncture works; response within the brain might be the most critical part. Once we better understand how acupuncture works to relieve pain, we can optimise this therapy to provide effective, non-pharmacological care for many more chronic-pain patients.

How time stopped circling and percolating and started running on tracks

As Ogle reminds us, it took nearly 44 years after the introduction of the Indian Standard Time in 1906 for the Bombay Municipal Corporation to finally agree to abandon its adherence to Bombay time, and thus bring to an end the now little-remembered ‘Battle of the Clocks’.

The ever-watchful gaze of the algorithms run by the state, corporations and technologies that document all our actions seem to bet on this conceit — given enough time under observation, their learning algorithms will have us figured. Time becomes the fire in which the steel of surveillance is sharpened. Amid all these vast imperious forces jostling to govern and influence us, we live our lives as if we are immortal. The occasional quests of freedoms we embark upon to recover our elusive selves remains our only way of attesting to our presence on this Earth. All the rest, we know deep within, will eventually surrender to time.

AI

OpenSpiel: A Framework for Reinforcement Learning in Games

OpenSpiel is a collection of environments and algorithms for research in general reinforcement learning and search/planning in games. OpenSpiel supports n-player (single- and multi- agent) zero-sum, cooperative and general-sum, one-shot and sequential, strictly turn-taking and simultaneous-move, perfect and imperfect information games, as well as traditional multiagent environments such as (partially- and fully- observable) grid worlds and social dilemmas.

This is something I really really need to look at a bit closer. Paper here

Blockchain

Blockchain Business Use Cases By Industry

  1. KYC
  2. serve unbanked
  3. cross-border transaction
  4. compliance — regulatory reporting

Some blockchain initiatives begin at the innovation lab of a larger organization or are incubated as part of a larger program, and if they’re successful, eventually spin out to become an independent business. In other instances, the use case/ business model warrants the creation of a new startup from the get-go. These startups can easily become a dominant provider in a small market since they’re agile and not bogged down by traditional IT.

The State of Stablecoins, 2019

Just a quick reminder of the types:

Fiat-collateralized

IOU “Centralized” stablecoins / backed by fiat currencies or commodities collateral-backed. Examples of these include Tether, USD Coin, Gemini, and Digix.

Crypto-collateralized

IOU “Decentralized” stablecoins / backed by crypto and/or multiple assets, collateral-backed. Examples of this kind of stablecoin include MakerDAO, Steem, and Alchemint.

Non-collateralized

Seigniorage Shares / Decentralized Bank / Algorithmic stabilization mechanisms. Examples of this kind of stablecoin include Terra, Ampleforth, and Element Zero

Culture

Remote company culture guide for the Slack generation

Pretty good read, interesting to delve into. Ultimately trying to upsell you into their product which is employee onboarding.

Documenting Decisions in a Remote Team

1. Clarify the decision

2. Decide & Share the Decision Makers

use RACI

3. Share That the Decision will Happen

4. Document the Decision

Don’t share in only one medium

Data

Implementing multi-class text classification with Doc2Vec

For complex text classification algorithms, the BOW would not be suitable as it lacks the capability to capture the semantics and syntactic order of words in the text. Thus using them as feature input to machine learning algorithm will not yield significant performance. Doc2Vec on the other hand is able to detect the relationships among words and understands the semantics of the text.

I’m just surprised how much times have changed. This was exactly what we did to classify fonts. So it was basically Doc2Vec2Font

Graph Analytics — Introduction and Concepts of Centrality

  1. Degree centrality — number of in vs out degrees
  2. Closeness centrality — shortest path between nodes
  3. Betweenness centrality — importance of a node in a network based on how many times it occurs in shortest path
  4. Eigen vector centrality — importance of node in graph as function of the importance of its neighbors

How I Spent My Time As Product Data Scientist

Development

How to do a code review

The pages in this section contain recommendations on the best way to do code reviews, based on long experience. All together they represent one complete document, broken up into many separate sections. You don’t have to read them all, but many people have found it very helpful to themselves and their team to read the entire set.

Good to serve as reference

Deploying a Machine Learning Model as a REST API

This was only a very simple example of building a Flask REST API for a sentiment classifier. The same process can be applied to other machine learning or deep learning models once you have trained and saved them.

The tutorial is basically about deploying REST APIs.

Stevey’s Google Platforms Rant

the insight that he can’t build one product and have it be right for everyone. But it doesn’t matter, because he gets it. There’s actually a formal name for this phenomenon. It’s called Accessibility, and it’s the most important thing in the computing world.

The. Most. Important. Thing.

But I’ll argue that Accessibility is actually more important than Security because dialing Accessibility to zero means you have no product at all, whereas dialing Security to zero can still get you a reasonably successful product such as the Playstation Network.

The Golden Rule of Platforms, “Eat Your Own Dogfood”, can be rephrased as “Start with a Platform, and Then Use it for Everything.” You can’t just bolt it on later. Certainly not easily at any rate — ask anyone who worked on platformizing MS Office. Or anyone who worked on platformizing Amazon. If you delay it, it’ll be ten times as much work as just doing it correctly up front. You can’t cheat. You can’t have secret back doors for internal apps to get special priority access, not for ANY reason. You need to solve the hard problems up front.

How to use Chart.js

Quick demonstration with line charts, how to add colours, multiple data sets, mixed chart

Boost your JavaScript Debugging Skills With These Console Tricks

console.log — logs what is done

console.count — outputs number of times it has been called with the label logged

function sayHello(name) {
console.count()
console.log(name)
}

console.warn — prints what is inside .warn("string") when something goes wrong

function sayHello(name) {  
if(!name) {
console.warn("No name given")
}
}

console.table — display elements in the array

console.group — create a new nested block of console statements

How to learn D3.js

Superb! Uses D3 to model how D3 looks. It’s modules etc

How Developers Think: A Walkthrough of the Planning and Design Behind a Simple Web App

  1. What does it do
  2. How does it look
  3. Where do you put it
  4. How does the code look
  5. What can go wrong

Life Optimisation

Six Brain Hacks To Learn Anything Faster

  1. Teach someone else
  2. Learn in short bursts of time
  3. Handwrite notes
  4. Mental spacing
  5. Study nap
  6. Change it up when you practice so you master it faster

How to disagree

Name-calling > Ad hominem > Responding to tone > Contradictions > Counterargument > Refute the central point (best)

But the greatest benefit of disagreeing well is not just that it will make conversations better, but that it will make the people who have them happier. If you study conversations, you find there is a lot more meanness down in DH1 than up in DH6. You don’t have to be mean when you have a real point to make. In fact, you don’t want to. If you have something real to say, being mean just gets in the way.

If moving up the disagreement hierarchy makes people less mean, that will make most of them happier. Most people don’t really enjoy being mean; they do it because they can’t help it.

Product

Why this Google product manager left to work on criminal justice reform

A large amount of criminal justice data exists, but it’s difficult for decision-makers to access. Researchers on staff may spend their time producing a single annual report. Jacoby says that the startup talked with one state that had 90 programs but only had the capacity to evaluate three in a year. “So in 30 years, we’ll know what was working in 2019,” she says. “They have the data, they know the questions they want answered, they just don’t have the capacity to answer their questions in any routine or deep way — and that’s exactly the place where tech can help.” Many tech companies do already work in the system, “but you can make a lot of money as a for-profit tech company without decarcerating at all,” she says. “The incentives in this space are not generally aligned.” That’s one reason the startup decided to launch as a nonprofit.

When Recidiviz first begins working with a state, they hold a design sprint with the state corrections department, learning what information the team has now, what they want to have, and how new data would help them make decisions differently.

The Product Management Capability Self Assessment

The six key attributes of great product managers

  • Product and design insight

Focusing questions: Are you an expert on your customers/users? Do you empathise with them? Do you constantly put them at the centre of what you do? Do you understand how your product works? Do you have a strong understanding of product risks of new features and how to address them? Do you manage risks up-front before you decide what to build?

  • Impact and results

Focusing questions — Do you focus on outcomes? Has your product meaningfully improved since your took over? What have you delivered? What impact have you had? What is your current goal for your product? How will you know if you achieved it? Why this goal?

  • Strategy and leadership

Focusing questions — Do you have a vision for your product? Have you shared it with your team? Are motivated and inspired to deliver it? Does the wider business understand and support your vision? Do you team look to you for leadership, especially in stressful situations?

  • Analytical fluency

Focusing questions — Can you get the numbers you need? Can you define metrics which are meaningful for your product? Can you spot the difference between a vanity and a meaningful metric? Can you analyse your a/b tests and get meaningful results? Are you able to utilise metrics to drive product optimisations? Can you interpret and extract insights from experiments and tests?

  • Technical literacy

Focusing questions — Can you talk in a language your developers understand and follow the response you get back? Do you understand how your product fits together? Can you meaningfully challenge your developers suggestions and spark productive discussions to drive better implementation? Can you roughly anticipate the complexity of building a feature you’re suggesting? Do you understand what your role is in the development process?

  • Communication and collaboration

Focusing questions — Are you an effective communicator? Can you simply explain what you are working on and its business impact? Do you have collaborative relationships with your developers and designers? Do your stakeholders in the business — eg. marketing, sales, finance — see you as a partner? Have you established mutual respect and trust with your stakeholders?

A beginner’s guide to user journey mapping

How:

  1. Work from user personas
  2. Research
  3. Map the journey

Why:

Identify pain points to help solve them. This is an opportunity to upsell them too.

How to turn your top help articles into in-app guides

Asana — succinct walkthrough

Github — tooltips

1. Identify speed bumps and bottlenecks

The first thing you need to know is where users are getting stuck. Review your help page doc views and support ticket history to see where users have the greatest demand for information.

2. Pick your battles

Once you have identified the opportunities, narrow your sights down to the top 1 to 3 items. Bonus points if you can cross check your list of friction points against the key actions that contribute to your product’s activation metric and find opportunities that fall into both categories.

3. Choose your locations

Just like in real estate, the success of your in-app guides has a lot to do with location, location, location. Consult your user journey map to plot out where and when each in-app guide will be most relevant and helpful to the user.

4. Create your assets

Now comes the fun part — creating your actual tooltip, walkthrough, or other messaging sequence. Remember to balance thoroughness with brevity, and don’t be afraid to bring a little personality into the mix.

5. Measure, iterate, repeat

Like other aspects of your product, in-app guides are rarely — if ever — a once-and-done thing. Measure their performance, and be prepared to iterate. This might mean updating copy or changing placement or switching up the overall flow. Be sure your product and development teams make room on their roadmaps to revisit your in-app guides as needed — or consider investing in a codeless third-party solution like Appcues that lets you publish tooltips, hotspots, and modals in a matter of minutes (not months).

Tools

Outcode

So before any coding starts we, ensure:

  • you understand the languages and frameworks you’ll have your project built in
  • you have the necessary designs for the dev to get started. If not, we put you in touch with one of our designers

We then match you with the dev we believe is the best fit for your task. All devs come from the small, talented and diverse community of freelancers we’re building. During the project, you’ll comunicate directly with the dev, but we’re here if you need anything.

Business Strategies and Frameworks (Part 1)

--

--

No responses yet