Dixon: 80% accurate ML models take a weekend; the rest takes decades
Chris Dixon · Fireside Chat: Chris Dixon, General Partner at Andreessen Horowitz (FirstMark's Data Driven) · Jun 8, 2018 · at 14:31
Chris Dixon, General Partner at Andreessen Horowitz, discusses the 80/20 problem in machine learning startup strategy at Data Driven NYC.
0:00 / 0:15exact quote · 15.4s
▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark
“You can sort of, like, you know, you can download TensorFlow, download some data sets, and over the weekend, probably come up with, you know, if you're a good programmer, come up with something that can do, like, 80% accuracy of whatever, let's say OCR or something, right? But then the last 20% takes you know, a decade if not a lifetime or something, right?”
quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →
More from Chris Dixon
Dixon: AI winners will be determined by data, not algorithms
“Is who kind of wins in AI, and I think that will probably come down to who has the best data, because the algorithms are, you know, all published and out there, and the frameworks are out there, and so it's not really, that's not really gonna be a differentiat…”
Dixon: Crypto networks are community-owned digital services, not corporate ones
“The way I think of crypto networks, as I call them, is, is they are digital services, and that's, I mean, in the broadest sense of any kind of digital service that are owned and operated by communities as opposed to being owned and operated by companies.”
Dixon: Centralized platforms predictably turn hostile to third-party developers over time
“What tends to happen with these big platforms is they start off being very open to, kind of, third parties to startups. Like, if you look at Facebook early on, they embraced Zynga, and they embraced media companies and things like this, and then as they got bi…”
Chris Dixon: The best VC strategy is following smart developers
“The thing that's worked for me as an investor is to not try to outsmart, ah, really smart developers. Just sort of let them show me where to go and follow them.”
Dixon: Musicians will make money selling digital virtual goods
“And so I think you're going to see musicians who now make money by selling digital virtual goods as an example, which is not a model today, and that's like a very interesting new model as an example.”
Dixon: Betting against Ethereum means betting against top developer talent
“If you want to bet against Ethereum as an example, like you're effectively betting against, you know, I don't know, like a lot of really, really smart engineers, and I just, I think it's a, You know, I'll take the other side of the bed.”