Hello!
Hello!
I’m Maggie
Hello!
I’m Maggie
Nomially,
“Staff Research Engineer” at GitHub Next
Hello!
I’m Maggie
More accurately,
design engineer
GitHub Next
The R&D labs team
Where is the industry is headed? What should GitHub do next?
GitHub Next
The R&D labs team
Where is the industry is headed? What should GitHub do next?
Other than fix uptime.
No, I don’t work on infra.
Plz don’t ask me.
My past titles
Illustrator
Art Director
UI/UX Designer
Product Designer
Design Engineer
A very quick pre-amble
What is a design engineer?
It is not
Making fancy microinteractions and posting them to twitter
It is not
Making fancy microinteractions and posting them to twitter
That is not a job.
My preferred definition
A designer deeply involved in the technical craft & material of software
You help make decisions about functionality and architecture, in code
Design engineers should:
State, data flow, databases, auth, APIs, algorithms, AI models, servers, latency, etc.
Fortune Telling
and Speculative Design
For fun and profit
We’re in a weird time. Everyone’s got whiplash from the speed of progress.
Hard to understand today’s reality, let alone figure out how to react.
Why even attempt to predict the future? (clearly a fool’s errand)
We can’t just give up or opt out because things are moving too fast. We need to think like fortune tellers and attempt to crystal ball.
Design engineers are the perfect people to figure out the shape of the future.
Uniquely positioned to prototype theories of the future in convincing ways.
Agents can “draw the rest of the owl” enough for you to make proof-of-concepts without full-stack skills (although they help).
What do I know about this?
I’ve been right and early on a few future predictions
How GitHub Next does speculative design
Labs work looks glamorous from the outside, and some of it is.
Play around and make prototypes all day
Without having to worry about shipping to production
Play around and make prototypes all day
A little, but sadly not really.
We do make lots of prototypes
But, we also need to:
obviously
The actual job
The actual job
is actually hard
The political struggles are real...
The political struggles are real...
Practical tools for speculative design
Ways to think harder, spot trends early, and properly explore what they mean
The tools
The futures cone
The futures cone
The futures cone
Weak signals
Early signs of change that are plausible, novel, significant and timely.
Collect them, pay attention to them, and look for themes across them.
Weak signals
Toys and prototypes, hot takes, product announcements, arXiv papers, newsletters, podcasts, VC cash flow, startups, economic trends, your own experiments and experience
Finally get rewarded for being chronically online
(Sorry, but for AI research work, yes you do need to be on X and Hacker News)
Weak signals
Toys and prototypes, hot takes, product announcements, arXiv papers, newsletters, podcasts, VC cash flow, startups, economic trends, your own experiments and experience
Finally get rewarded for being chronically online
(Sorry, but for AI research work, yes you do need to be on X and Hacker News)
Weak signals
Toys and prototypes, hot takes, product announcements, arXiv papers, newsletters, podcasts, VC cash flow, startups, economic trends, your own experiments and experience
Finally get rewarded for being chronically online
(Sorry, but for AI research work, yes you do need to be on X and Hacker News)
Weak signals
Toys and prototypes, hot takes, product announcements, arXiv papers, newsletters, podcasts, VC cash flow, startups, economic trends, your own experiments and experience
Finally get rewarded for being chronically online
(Sorry, but for AI research work, yes you do need to be on X and Hacker News)
What can we conclude?
What can we conclude?
Second and third order effects
Take the observation and ask what happens next and next and next
Second and third order effects
Take the observation and ask what happens next and next and next
Sandboxed machines and always-on proactive agents
OpenClaw, Muse, etc.
Sandboxed machines and always-on proactive agents
OpenClaw, Muse, etc.
Play with your ideas
Doing (cheesy) structured exercises to think deeply about your theories of the future and understand your assumptions.
→ the agents prompt us
We build the software with agents
→ the software builds itself
Can the agent share memories within a team or family?
Can the agent tactfully forget sensitive facts or details?
Could an agent choose to delete evidence to protect its human?
Artefacts from the future
Make provocative artefacts that make your potential future feel tangible and real
Also called “diegetic prototypes”
Artefacts from the future
Make provocative artefacts that make your potential future feel tangible and real
Cricket Crunch by Julian Bleeker of the Near Future Lab, 2020
What artefact would make this future evocative?
Playful, provocative, interesting, and specific, over “realistic”
Can be images, objects, adverts, specific interface patterns: changelogs, code review, software updates, errors, warnings.
Failure states often reveal more than happy paths
Different prototypes for different jobs (and audiences)
Moderate, near-future, “realistic” artefacts
Crazy, far-future, speculative, “impractical” artefacts
Make your theories public and concrete
Use writing and conference talks to build theories. Write to clarify.
Tell people your theories. If it comes true, you get cred. If it doens’t no one remembers.
Back it up with data. The pressure of telling people and public debate will make you research it properly.
GitHub Next is hiring!
Design engineers / very senior front-end perfectionist with good design taste. Please come talk to me or message us after:
@idangazit
@mappletons
Thank you.
maggieappleton.com
githubnext.com