StarZero All shows Try the podcast clipper →

Sep 28, 2017 · 25m · mad

A Fresh Approach to Technical Computing // Viral Shah & Stefan Karpinski, Julia Computing

Viral Shah · 11m spoken Stefan Karpinski · 9m spoken Matt Turck · 44s spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

Julia Computing co-founders Stefan Karpinski and Viral Shah present Julia, a dynamic high-performance programming language designed to solve the 'Two Language Problem' in technical computing. They showcase Julia's architecture, ecosystem growth, real-world enterprise applications, and address business models and migration in a Q&A session.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 3.5% of the talking time here. How this is scored →

Matt as informed peer 0.4 Guest teaching 3.0 Guest disagreement 1.2 Matt pushing back 0.4
05100:0010:0020:000:14–2:54 · Matt as informed peer 0/10 Stefan Karpinski on Data Science in 2009 & Julia's Origin This is a solo presentation segment by guest Stefan Karpinski without host participation. Karpinski humorously describes his overly complex 'Rube Goldberg' data science stack in 2009 that led to Julia's creation.2:54–7:39 · Matt as informed peer 0/10 The Two Language Problem & Standard Compromises Stefan presents the theoretical concept of the 'two language problem' and Osterhout's dichotomy without host interaction. He outlines the performance and productivity drawbacks of sandwiching high-level code with C.7:39–11:28 · Matt as informed peer 0/10 The Julian Unification, Benchmarks, and Productivity Stefan completes the presentation portion on Julia's benchmark speed versus developer productivity before handing off to Viral Shah. The host is absent during this monologue segment.11:28–16:01 · Matt as informed peer 0/10 Real-World Industry Applications and Case Studies Viral Shah presents real-world applications in astronomy, banking, and aviation. The host does not speak or engage during this monologue presentation.16:01–25:43 · Matt as informed peer 2/10 Closing Remarks and Paul Graham Quote Host Matt Turck joins for Q&A along with audience members, asking about business models and community building. The guests educate the host and audience on open-source monetization realities and why automated code translation is impractical.0:14–2:54 · Guest teaching 2/10 Stefan Karpinski on Data Science in 2009 & Julia's Origin This is a solo presentation segment by guest Stefan Karpinski without host participation. Karpinski humorously describes his overly complex 'Rube Goldberg' data science stack in 2009 that led to Julia's creation.2:54–7:39 · Guest teaching 3/10 The Two Language Problem & Standard Compromises Stefan presents the theoretical concept of the 'two language problem' and Osterhout's dichotomy without host interaction. He outlines the performance and productivity drawbacks of sandwiching high-level code with C.7:39–11:28 · Guest teaching 3/10 The Julian Unification, Benchmarks, and Productivity Stefan completes the presentation portion on Julia's benchmark speed versus developer productivity before handing off to Viral Shah. The host is absent during this monologue segment.11:28–16:01 · Guest teaching 3/10 Real-World Industry Applications and Case Studies Viral Shah presents real-world applications in astronomy, banking, and aviation. The host does not speak or engage during this monologue presentation.16:01–25:43 · Guest teaching 4/10 Closing Remarks and Paul Graham Quote Host Matt Turck joins for Q&A along with audience members, asking about business models and community building. The guests educate the host and audience on open-source monetization realities and why automated code translation is impractical.0:14–2:54 · Guest disagreement 1/10 Stefan Karpinski on Data Science in 2009 & Julia's Origin This is a solo presentation segment by guest Stefan Karpinski without host participation. Karpinski humorously describes his overly complex 'Rube Goldberg' data science stack in 2009 that led to Julia's creation.2:54–7:39 · Guest disagreement 1/10 The Two Language Problem & Standard Compromises Stefan presents the theoretical concept of the 'two language problem' and Osterhout's dichotomy without host interaction. He outlines the performance and productivity drawbacks of sandwiching high-level code with C.7:39–11:28 · Guest disagreement 1/10 The Julian Unification, Benchmarks, and Productivity Stefan completes the presentation portion on Julia's benchmark speed versus developer productivity before handing off to Viral Shah. The host is absent during this monologue segment.11:28–16:01 · Guest disagreement 1/10 Real-World Industry Applications and Case Studies Viral Shah presents real-world applications in astronomy, banking, and aviation. The host does not speak or engage during this monologue presentation.16:01–25:43 · Guest disagreement 2/10 Closing Remarks and Paul Graham Quote Host Matt Turck joins for Q&A along with audience members, asking about business models and community building. The guests educate the host and audience on open-source monetization realities and why automated code translation is impractical.0:14–2:54 · Matt pushing back 0/10 Stefan Karpinski on Data Science in 2009 & Julia's Origin This is a solo presentation segment by guest Stefan Karpinski without host participation. Karpinski humorously describes his overly complex 'Rube Goldberg' data science stack in 2009 that led to Julia's creation.2:54–7:39 · Matt pushing back 0/10 The Two Language Problem & Standard Compromises Stefan presents the theoretical concept of the 'two language problem' and Osterhout's dichotomy without host interaction. He outlines the performance and productivity drawbacks of sandwiching high-level code with C.7:39–11:28 · Matt pushing back 0/10 The Julian Unification, Benchmarks, and Productivity Stefan completes the presentation portion on Julia's benchmark speed versus developer productivity before handing off to Viral Shah. The host is absent during this monologue segment.11:28–16:01 · Matt pushing back 0/10 Real-World Industry Applications and Case Studies Viral Shah presents real-world applications in astronomy, banking, and aviation. The host does not speak or engage during this monologue presentation.16:01–25:43 · Matt pushing back 2/10 Closing Remarks and Paul Graham Quote Host Matt Turck joins for Q&A along with audience members, asking about business models and community building. The guests educate the host and audience on open-source monetization realities and why automated code translation is impractical.

speaking balance: gold is Matt, purple is the guest (3 minute bins)

0:00 · Matt 0% · guest 100%0:00 · Matt 0% · guest 100%3:00 · Matt 0% · guest 100%3:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%9:00 · Matt 0% · guest 100%9:00 · Matt 0% · guest 100%12:00 · Matt 0% · guest 100%12:00 · Matt 0% · guest 100%15:00 · Matt 13.2% · guest 86.8%15:00 · Matt 13.2% · guest 86.8%18:00 · Matt 11.9% · guest 88.1%18:00 · Matt 11.9% · guest 88.1%21:00 · Matt 6.4% · guest 93.6%21:00 · Matt 6.4% · guest 93.6%24:00 · Matt 1.5% · guest 98.5%24:00 · Matt 1.5% · guest 98.5%
Sharpest disagreement ▶ 24:50 Stefan rejects automated translation premising

Stefan firmly dismisses the audience premise that automatic translation tools from Python/R to Julia are feasible, explaining that corner cases make automated transliteration fundamentally flawed.

 Hardest push from Matt ▶ 18:04 Matt questions commercial readiness

Matt presses the guest on product availability, pointing out that despite the underlying project's age, the company itself is young and may not have products ready.

 Biggest teaching moment ▶ 16:48 Viral educates on open-source business models

Viral breaks down why traditional open-source monetization tactics like consulting or open-core fail, explaining that enterprise 'plumbing' and cloud infrastructure are the true paths to scale.

 Matt holds his own ▶ 16:38 Matt frames open-source commercialization challenge

Matt demonstrates industry knowledge by noting how rare commercial success is for open-source foundation languages and prompting the guest to justify their strategy against precedent.
the scores for every segment, with the reasoning behind each
ChapterTopicMatt as informed peerGuest teachingGuest disagreementMatt pushing backWhy
Stefan Karpinski on Data Science in 2009 & Julia's Origin 0210 This is a solo presentation segment by guest Stefan Karpinski without host participation. Karpinski humorously describes his overly complex 'Rube Goldberg' data science stack in 2009 that led to Julia's creation.
The Two Language Problem & Standard Compromises 0310 Stefan presents the theoretical concept of the 'two language problem' and Osterhout's dichotomy without host interaction. He outlines the performance and productivity drawbacks of sandwiching high-level code with C.
The Julian Unification, Benchmarks, and Productivity 0310 Stefan completes the presentation portion on Julia's benchmark speed versus developer productivity before handing off to Viral Shah. The host is absent during this monologue segment.
Real-World Industry Applications and Case Studies 0310 Viral Shah presents real-world applications in astronomy, banking, and aviation. The host does not speak or engage during this monologue presentation.
Closing Remarks and Paul Graham Quote 2422 Host Matt Turck joins for Q&A along with audience members, asking about business models and community building. The guests educate the host and audience on open-source monetization realities and why automated code translation is impractical.

Statements from this episode (13)

Opinion
Karpinski: R produces the best plots and leads statistical analysis
“I think it makes the best plots and it's, you know, the undisputed king in, in statistical analysis.”
Stefan Karpinski Sep 28, 2017 ▶ 1:21
Assertion Partly supported
Karpinski: Julia surpassed NumPy in contributors within one year
“Even when we were one year into being a public project, we already had more contributors and developers than NumPy, and I could not understand how that was possible, because I was like, we're this very young project, this was, you know, back in 2013”
Stefan Karpinski Sep 28, 2017 ▶ 5:52
Insight
Karpinski: Alternating language layers in AI frameworks prevents compiler optimizations
“They have what I've, I would describe as a sandwich problem, which is that you end up sandwiching a lot of system code with user code, and then, like, adding more and more layers of that, and as you've sandwiched, like, you know, seven or eight layers of that,…”
Stefan Karpinski Sep 28, 2017 ▶ 6:39
Assertion Not checkable as stated
Karpinski: Julia's real-world execution speed stays within 2x of C
“The real life, life experience of what people have done, you know, in like, you know, years of programming in Julia now, is that they tend to see that Julia is within one to two of C.”
Stefan Karpinski Sep 28, 2017 ▶ 8:11
Assertion Not checkable as stated
Shah: Julia programming language has surpassed 1.2 million downloads
“We've had over 1.2 million downloads until today, even though the research is anchored at MIT.”
Viral Shah Sep 28, 2017 ▶ 10:29
Assertion Supported
Shah: The New York Fed uses Julia to model the US economy
“The New York Fed uses Julia to model the US economy”
Viral Shah Sep 28, 2017 ▶ 11:42
Assertion Not checkable as stated
Shah: Julia is the only non-C language with native CUDA codegen
“Apart from CUDA and C, the toolkits that NVIDIA puts out, Julia is the only other, you know, widely used language that has native CUDA code gen.”
Viral Shah Sep 28, 2017 ▶ 12:14
Assertion Partly supported
Shah: Aviva uses Julia for all Solvency II regulatory reporting
“Aviva, one of the largest insurers in Europe, uses Julia for all of their regulatory reporting around solvency too.”
Viral Shah Sep 28, 2017 ▶ 14:03
Assertion Contradicted
Shah: Central Bank of Brazil manages $1 trillion assets using Julia
“The Brazilian National Bank uses Julia to manage all of its trillion dollars in assets.”
Viral Shah Sep 28, 2017 ▶ 14:12
Opinion
Shah: Julia's JuMP framework beats any commercial optimization system
“And it's by far better than any commercial system out there”
Viral Shah Sep 28, 2017 ▶ 15:28
Insight
Shah: Consulting is not a viable open-source business model
“Consulting is a way to build an open source business, but it's obviously not because you don't want your core devs to be consulting when you really need to be building the product.”
Viral Shah Sep 28, 2017 ▶ 16:57
Insight
Shah: Open-core monetization fails by alienating the user community
“Open core models don't work very well because the community that was, you know sort of going with you starts hating you because all the good stuff is not there anymore.”
Viral Shah Sep 28, 2017 ▶ 17:07
Insight
Karpinski: Automatic dynamic code translation fails due to edge cases
“The trouble is when you get into automatic translation, it's always the corner cases that kill you. And the corner cases that kill you with automatic translation are actually exactly the reasons why you can't just sit down and be like, oh, let's just make Pyth…”
Stefan Karpinski Sep 28, 2017 ▶ 24:59
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 400 conversations transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.

Try the podcast clipper Learn more