Blog

Feedback loops and the limits of responsibility

Feedback loops in machine learning are often presented as a problem that arises in particular domains, with familiar examples in recommendation systems and predictive policing. These examples are usually treated as special cases. In this piece, I argue the opposite: feedback loops are the default condition of deployed machine learning systems, not a pathological edge case. Feedback loops are structurally inevitable once models are deployed into the world because of how we frame, evaluate, and govern these systems.

Gabrielle Josling

Access rights in practice: reflections from 60 data access requests

In Australia, you have the right to access the personal information an organisation holds about you under Australian Privacy Principle 12 (APP 12). Over the past year, I’ve made 60 data access requests to a range of organisations, including charities, major retailers, and data brokers. The process revealed as much about organisational governance and privacy maturity as it did about the data itself.

Gabrielle Josling

Can disabled people…? What autocomplete reveals about views on disability

Inspired by Safiya Umoja Noble’s Algorithms of Oppression, this post explores a neglected area of autocomplete research. While stereotyping in search suggestions has been studied for race, gender, and other attributes, disability is largely absent from the literature. I report the results of a small audit of disability-related autocomplete queries across major search platforms.

Gabrielle Josling

A taxonomy of awkward conversations with stakeholders

This post is about different categories of difficult conversations with stakeholders that you will almost inevitably encounter as a data scientist. I think of it as more of a taxonomy (and perhaps commiseration) than advice. Data scientists sit in an awkward place inside organisations. We’re expected to deliver authority, certainty, and progress, while working with uncertainty, partial information, and value judgements that rarely belong to data science alone. Many of the most uncomfortable conversations arise from that tension.

Gabrielle Josling

Data ethics is your job

All too often in data science projects, data ethics is treated as someone else’s responsibility. It sits with legal, governance, product, or a committee that meets occasionally and far away from the day-to-day work. If it comes up at all, it’s often well after the key decisions have been made about what gets built. This post argues that data ethics is not an optional extra. Data scientists are often uniquely positioned to identify ethical risks and potential harms, and that position comes with an obligation to take them seriously.

Gabrielle Josling

Measuring how long things take is harder than it seems

Measuring how long a process takes seems easy: start, stop, average. So it should disturb you to find out there’s an entire subfield of statistics devoted to it. This post looks at why time-to-event data is trickier than it seems and the mistakes people commonly make.

Gabrielle Josling

The case against propensity modelling

Propensity models look for patterns in past data to estimate how likely someone is to do something (buy, cancel, or upgrade). It’s a deceptively simple idea: train a model on who acted before and use it to predict who will act next. It sounds powerful, and in theory it should let you focus your efforts where they’ll have the most impact. In practice, the question propensity modelling answers is often the wrong one.

Gabrielle Josling

Everything breaks eventually: data science lessons I’ve learned the hard way

A few years in data science will give you strong opinions about strange things: Microsoft Excel, daylight saving time, unit tests. The beauty of this job is that things can always go wrong in new and interesting ways, but I’ve gradually accumulated some hard-won instincts about what’s most likely to break first. These are a few of the lessons I’ve learned the hard way on the technical side of the job. The human side deserves its own list (or possibly a thesis).

Gabrielle Josling

The hidden work of data science projects

You’ve probably heard the phrase “80% of data science is data cleaning.” It’s not totally wrong, but it is incomplete. While I agree that data science often involves surprisingly little modelling, the 80% line doesn’t do justice to the full range of non-modelling work that makes data projects succeed. These are the things that tend to go unnoticed and unappreciated if everything goes well — but skip them at your peril.

Gabrielle Josling

ScoMo’s last hurrah?

When I started collecting and analysing Australian Prime Minister Scott Morrison’s speeches and interview transcripts over a year ago for a blog post, I hardly believed he would still be Prime Minister come the next election. With the election today, this is very possibly Scott Morrison’s last day as PM. In honour of the occasion, this is a special edition post about ScoMo’s language use during the election campaign.

Gabrielle Josling

Automated web scraping using GitLab CI/CD

When working on data projects that rely on regularly updated sources, it’s easy to end up with a manual refresh process — running a series of scripts locally, saving the results, and uploading the new data wherever it’s needed. That approach works fine at first, but quickly becomes repetitive and error-prone as the project grows. This post describes how I used GitLab CI/CD and AWS S3 to automate a set of R scripts that collect, clean, and process data on a schedule.

Gabrielle Josling

The Prime Minister’s speech: A textual analysis

Since becoming Prime Minister in 2018, Scott Morrison has spoken publicly almost every day. Transcripts of all speeches, interviews, and press conferences are available on the Prime Minister’s website, which means we can analyse ScoMo’s language over this whole period (August 2018 - March 2021).

Gabrielle Josling

UI/UX and data science

In this post I’ll talk about what I learned from two courses I recently took on Coursera: Visual Elements of User Interface Design and UX Design Fundamentals (both taught by Michael Worthington from the California Institute of the Arts).

Gabrielle Josling

Unsupervised learning and lessons from the science of personality testing

After completing a personality test as part of a job application, I recently became interested in the science and statistics of personality testing. This is part of the field of psychometrics, which is concerned with the measurement of mental traits and aptitudes. Psychometrics is a fascinating field in itself, but the more I read the more I started to see parallels between the challenges of personality testing and those of unsupervised learning. In this post I’ll discuss some of the things that make personality testing difficult and how the the field of personality testing deals with these challenges. While personality testing does have some unique challenges, often there are lessons to be learned from the field that can be applied to unsupervised learning problems more generally.

Gabrielle Josling

A deep dive into short circuit evaluation

One consequence of not having ever learned programming in any systematic way is that sometimes I come across a very unexpected behaviour that I don’t really have the language to describe (or more problematically, to Google). Recently when this happened I was led down a very interesting rabbit hole to the idea of short circuit evaluation. Importantly, I learned that my mental model for how logical operators work was not quite right.

Gabrielle Josling