Moving to Substack
When I published my first story on Medium — a New Year’s post on Jan 3 2017 — it did great numbers. We got 55K views, according to Medium (and probably more in practice via reader apps), I got a lot of positive feedback, and it ranked #1 on Google for GANs (generative adversarial networks) for quite some time. It probably still comes up in search to this day. Half of the official view count is from “google.com” — according to Medium.
GANs will change the world
It’s New Year’s 2017, so time to make predictions. Portfolio diversification has never been me, so I’ll make just one.
I had no presence on Medium at the time. Sure I had somewhat of a Twitter following — 2k followers, half of what I have now. And some prominent friends retweeted the piece. The writing and topic of course was also good, and timely.
But the Medium network helped a lot. There were “publications” on Medium back then — aggregators of good content, in this case on tech. They discovered and promoted good one-off pieces, curating the site, in what’s a good deal for everyone. [I’m a big fan of curation in general, and think this is sorely lacking in the NFT space, for example — a topic for a future post.]
I got a number of comments and highlights on the piece. Everything was awesome.
Future Medium pieces never did those same numbers. I didn’t write monthly, as I thought I would, more like a few times a year, and occasional book reviews. But everything that I posted — on machine learning, poker AI, baseball nerd stuff etc — did at least 5K views, got discovered, and ranked well on Google.
My baseball piece still gets traffic. Not just from people looking me up, for interviews and business purposes. It’s a good piece, although largely a link-back to baseball work I did in ~2009, and posted on the Blogger platform.
Machine Learning for Baseball ⚾ (my story)
A year after I left Google in late 2008 — I traveled around for a bit, got involved in a messy startup, etc — I was…
I kept doing annual AI review pieces, and those continued to do well into 2019. This long AI paper review piece got ~10K views, more than half of those from Google dot com, and still gets a trickle of traffic.
The Best Deep Natural Language Papers You Should Read — BERT, GPT-2 and looking forward
Transformer-type language models seem to be breaking NLP benchmarks every week. What do I need to know to keep up?
I love writing (although do so more rarely than I’d like, since my process takes forever), I’ve gotten a ton of positive feedback on my stuff, and my writing — since the Blogger days — has led to a ton of business and friendships.
There’s few better ways to connecting with creative and high achievement people than writing good (timely!) content and having people find it.
For me this has been a wonderful experience, although a distant second to poker, in terms of meeting interesting and exciting people. Something I’ve written about before. The gym, is third. Don’t try meeting people at the bar, folks. NGMI.
[An amazing experience I’ve had this year was getting to know Nate Silver. I’ve read his stuff for years, and been inspired to dive into baseball modeling because of his PECOTA projections. After admiring his work from afar for a long time, I felt like I knew the man when we were finally introduced in person through a mutual friend in Miami this spring. It has been great to be on both sides of that exchange. As I met my good friend and baseball legend Kyle Boddy, because he reached out to me over my own baseball blog, a decade ago.]
Medium’s problems have been well documented. I won’t bother to cite them here.
Over the years I’ve noticed my readership dwindling — and my follower count not growing. I’ve noticed the on-site SEO isn’t what it used to be — it used to be great! I never got much on-site Medium reader app discovery in the first place…
Something happened in 2020, and ever since that long late 2019 piece that did good numbers, nothing I’ve written has gotten 1K views. This also means the pieces don’t rank on Google. Maybe my writing over the past two years has not been as good. But either way people ain’t finding it.
Even my write-up on the crypto/NFT price prediction project I’ve been working on: DeepNFTValue.
Predicting CryptoPunk Prices: The Case for JPEGs
We trained a deep neural network to predict NFT prices.
The piece did pretty well on Twitter, and it has already generated good inbound traffic for me on from the DeFi and greater crypto community. My project is awesome and people are interested. The writeup helped. But I’m still not even getting 5–10% of the traffic I used to get for an ML explainer, a personal story, or a book review. BTW, here’s one of my favorite finance books (although this one actually got no views):
A Man for All Markets: 📚 Review
Note: for those who prefer listening to reading, Bloomberg’s Barry Ritholtz did a great interview with Ed Thorpe on…
Therefore, while I may still post on Medium occasionally, I am moving to Substack.
Which cryptos should I buy? The asset allocation model.
I often get the question - what cryptos should I buy? Not only at the poker table. But also, at the poker table... What…
My first piece is a typical (for me) explainer, talking through how I think about asset allocation and crypto, with a few stories and jokes thrown in. I cite heavily from Twitter. Always loved how well Medium integrates tweets into the stories — although that feature has also not really kept up with the evolution of Twitter formats. Substack does not render embedded links nearly as well as Medium, but it does handle tweets pretty good. Important given how much news and analysis happens in the Twitter streets, especially for the things I care and write about:
- Machine learning and deep learning
- Quant trading
- Nerd baseball
My content will continue to draw heavily on curating some of the alpha in this space, archiving it, summarizing it and dropping some 🌶️ opinions. I will also be focusing mostly on the intersection of ML and crypto, and explaining my DeepNFTValue NFT price prediction project.
I hope you follow me over to Substack. I don’t think I’ll be cross-posting content here, although maybe I should. The goal was never to get the most number of views, but to be seen, to be discovered, and in some magical cases interact with the readers (people usually find my DMs, but occasionally in public).
That is not happening on Medium for me, not any more. I know that others are still getting great numbers and discovery on this platform. For example Arthur Hayes awesome piece on NFTs.
JPEG The Jeweller
(Any views expressed in the below are the personal views of the author and should not form the basis for making…
However I feel that for occasional authors like myself, who don’t write often, and don’t have a massive name or tens of thousands of followers… Medium is NGMI.
Whatever happened at the company slowly over the past 3–5 years, and quickly over the last two… it’s just no longer working for me, in terms of distribution, indexing and discovery.
I hope that Medium allows this post to stay up. I am by no means a hater. Nor I am bitter about giving a lot to this site for free, or anything like that. I’ve gotten a ton in return — mostly in the friendships I’ve made from writing in public. Nothing lasts forever, business models evolve (or not) and most things come to an end.
I am not moving because of censorship, although this last paragraph makes me think how much the whole Substack vs Medium model is even viable for independent writers who might say something controversial once in a while. Neither platform is exactly the definition of Ben Thompson’s “Faceless publisher” — but Substack comes a lot closer in terms of letting you run your site and own your own audience.
When I first worked for a (student) newspaper, the job of a publisher seemed odd to me; as far as I and my editorial…
In the end of the day, many like me just want to write a good blog, with useful publication tools (remember, I started on Blogger 💩). We want good indexing, good out of the box SEO, view tracking, etc. We don’t want to be censored. Anything else on top is a bonus 🍒