138: Cloudflare's Credibility Problem, Elastic Q1, Salesforce + Mulesoft, Gmail is Broken, WSB Going Endemic, Stripe, Chrome's V8 Engine, MDMA vs PTSD, and 50MM Lenses
"A few flare ups here and there?"
I once heard a story about a man who uses a wheelchair.
When asked if it was difficult being confined, he responded, “I’m not confined to my wheelchair—I am liberated by it. If it wasn’t for my wheelchair, I would be bed-bound and never able to leave my house.”
This shift in perspective completely transformed how he lived each day.
—James Clear, Atomic Habits
📨 Gmail is crucial internet infrastructure. Gmail is broken. I wish Google would fix it.
The idea of creating sub-inboxes to automatically filter some stuff isn’t bad in itself.
Putting all those promotional emails (virtual “deal & coupon flyers”) from Uber, Walmart, and that supplement store where I once bought Omega 3 together in the “promotions” tab makes sense.
Problem is, that’s not all they’re sending there, and that’s what’s broken.
If I willingly sign up for a newsletter about semiconductors or mechanical keyboards, I want to read it. It should go to my inbox, not to the “promotions” tab with commercial emails — it just doesn’t belong there.
Yet Gmail sends a lot of newsletters there, and while plenty of people know the workarounds (add the newsletter source email, LibertyRPF@Substack.com, to your contacts, or create a rules-based filter that sends the newsletter to Inbox, etc), there are many many people who don’t know about this, and never see what they signed up for because it gets buried in Promotions, where they rarely go.
Even my ‘welcome’ email with instructions on how to do this may end up buried, so how can I tell people?
And if you never see something, you don’t become a fan of it, so there’s little motivation to jump through hoops to fix this issue. You may not even be aware there’s a problem. Out-of-sight, out-of-mind. It kills it in the egg.
Maybe there’s someone out there who would’ve been #1 fan of this newsletter for years, we would’ve become best-friends online, exchanged life-changing ideas, etc.
But we’ll never get to meet because after they signed up, they never saw the emails and forgot about it all…
Just thinking about it makes me sad.
This is a bit like if I bookmarked a website, but my web browser decided “nah, I won’t bookmark it, I’ll decide for you that you don’t get to see it”.
Or if I added a blog I wanted to follow to a RSS reader, but the app decided for me, “nah, I’ll just hide it from you”.
That’s broken behavior. Bad UX. User-hostile.
Gmail is a very very very large email provider, and by doing this, it’s hampering the flow of information to millions (anyone remember what Google’s mission is?) and thwarting its users’ intent (if you subscribe, you want to see it).
I really wish someone who works at Gmail would fix it.
🤔 I wrote before about how I think the human mind is a bit like a lobster trap 🦞 and ideas get in, but they’re very very hard to get out and there’s no easy mechanism to make people “unbelieve’ stuff.
This is why disinformation is so potent and why people will still automatically think of vaccines and autism together even after the original studies on this have been debunked and/or shown to be fraudulent, or why it took so long for people to update to the airborne transmission belief with Sars-CoV-2 even though we figured it out pretty early on.
I was thinking about psychedelics. I think the biggest thing that the people who do research and are developing therapies based on these compounds could do is rebrand them.
Find a new name for it. Ideally something fairly neutral and technical that could, over time, become associated with whatever results the field produces.
The word “psychedelics” has too much baggage, it’s no good anymore. It may work for me and you and people who are following the space fairly closely, but for most people who aren’t paying attention, the associations are too strong.
📝 Friend-of-the-show and supporter (💚🥃) Lillian Li (check out Chinese Characteristics, her great substack on “Chinese tech longform analysis”) emailed me something that made me see one of the good aspects of newsletters.
If you follow a NL writer on Twitter and they post about a lot of the same things, it’s more a drip-drip-dip, and even if you get to kind of the same place, you don’t get that big-xmas-present-to-open-all-at-once effect.
Lillian phrases it well: “the anticipation and payload are great. We might say that's why newsletters are so satisfying.”
They have downsides too, and the longer-form bias inherent to the medium can make them overwhelming… But hey! Life is trade-offs, and that’s why I encourage you not to feel like you have to read everything. From the welcome email:
Most editions won’t be linear or themed, and will be eclectic, so feel free to jump around and read only what grabs you based on sub-titles or images. No pressure to read everything!
💚 🥃 What happens if a tree doesn’t get water?
(we’re now at 3.8%, moving in the right direction!)
Investing & Business
Cloudflare: “You lose credibility”
I was read the transcript of a presentation by Cloudflare’s CFO recently, and he was talking about the company’s total addressable market (TAM — see, sometimes I make the effort to write out what the acronyms are to make it easier for those who aren’t as familiar with finance stuff), and how they’ve been expanding it with new products.
Then he talked about why they aren’t including their new Workers edge computing platform in any of their TAM numbers and projections:
it's really hard to quantify the edge computing market. The numbers get so big so fast. You tend to become -- lose credibility.
So basically, it looks like such a big market that they’re more-or-less embarrassed by the numbers that they’re coming up with and would rather just not include them.
Also on Workers:
we are not driving revenue at this point in time. It's much more about developer adoption, and making people use Workers for their specific use cases and products.
This is very smart, because Workers is a new 💫 platform 💫, and almost the only thing that matters at first is adoption.
Do devs learn the tools? Do they built things on it, have a good experience and so will keep building, and some of these things will grow into big things over time..? Is there a community forming that can share best practices among themselves and provide feedback to Cloudflare for better/faster product iteration?
These platforms have network effects, and network effects cut both ways, so velocity of adoption and development/improvement is what will later lead to the bucks, but in the meantime, it’s land-grab time, both out in the world, but also in dev mindshare.
Also, here’s some classic, textbook Clayton Christensen low-end disruption, climbing the ladder over time and challenging the existing players from below:
when Cloudflare got started now almost 11 years ago in October, it focused on the long tail of the market. So small and medium-sized businesses, developers, our first go-to-market models where what we call pay-as-you-go models where customers give a credit card, and we charge them $200 a month for the services we offered.
And only later came our enterprise go-to-market. Today, this is the majority of our business and our fastest growing business. [...] look at what we call our large enterprise customers, and we define those as customers that pay us more than $100,000 a year. This is now 50%, a little bit more than 50% of our revenue. [...]
Let's look at customers that give us more than $500,000 or even more than $1 million a year, the larger the cohort, the faster the growth. So our largest cohort, so $1 million-plus customers, has been growing north of 70% now consistently over last the 7 quarters [...]
if you go back over time, we seem to make a step function change in terms of customer size from $50,000 to $100,000 to the first $500,000 customer, the $1 million dollar customer to the first $10 million customer. So we are right at that brink currently.
Interesting point on churn, and the impact of multi-product bundles (something that Tom Rutledge at Charter is quite aware of!):
the uniqueness of the Cloudflare setup is really that installing a product or is really it's just a mouse click away in terms of product expansion. And we had the threshold of once we get to 4 customers -- 4 products per customer, then churn rates are coming down really significantly. We were at 75% -- or north of 70% of our customer using 4 products or more at the IPO. That number now has moved far beyond 80%. And now more than 70% of the customers are using 5 products on.
On their presence in China, and the relationship with JD (which I’m assuming they are colocating their equipment with?):
we started very early to develop our footprint in China in the relationship with Baidu that carried us to the current day. With Baidu, we had about slightly more than 20 cities in China that served that had our equipment and provided our services.
And that already made us very unique because we are able to offer this as one consistent network, one control plane regardless of whether you're doing business in China, outside of China or into china, one consistent offering, one network. [...]
We are evolving our approach there and moving from Baidu to a relationship with JD.
With JD, we are going to get to more than 50 cities in China by the end of this year. And an additional 100 cities, so in the 2 years thereafter.
Ok, one last point, because I try not to make this too long, but some companies just have more interesting stuff to talk about…
For us, COVID was an interesting period of time because when it all started, we saw more headwinds than tailwinds. Our traffic increased. Everybody was working from home. The internet became this important assistant for all of us to make it through the pandemic. So since we have a business model that is not charging for use, our costs went up, but we didn't pass that really on. [...]
some of the most exciting products we launched [during the pandemic] like Cloudflare for Teams and Cloudflare Access, we gave away for free
So you may assume that the pandemic was a tailwind to them like it was for many others, but from a financial point of view, it wasn’t. I still think over the longer-term, and on the net, the pandemic will have been positive for them because it accelerated many trends that they are riding on, but this stuff hasn’t really been reflected on the financials fully yet because of those headwinds. Interesting, no?
WallStreetBets going Endemic?
You know how some new viruses become endemic after a while, as Sars-CoV-2 is likely to do.
They stick around in the background and while they don’t cause as much trouble as when they first appear — because nobody had any immunity, or vaccines and therapies — they still simmer and flare up here and there once in a while.
It kinda feels like this is what is happening for the Reddit-WallStreetBets thing.
The first go-round, the media and financial ecosystem had no defenses, it was a novel pathogen, and for a few weeks earlier this year, it rapidly colonized everybody’s mindspace.
Nobody could talk or write about anything else, it seemed — felt like Bitcoin in 2017 when even Twitch streamers were talking about crypto with their chatrooms.
And now, a lot of the same stuff is happening… Gamestop is a $20bn market cap, AMC, a company that lost $149m in 2019, before the pandemic, is now a $30bn market cap and going up 100% in a day, and BlackBerry is spiking around 100% in a few days too…
This is making some headlines, and giving plenty of material to Matt Levine, but there’s definitely not the frenzy that we had the first go around.
So are WSB shenanigans becoming endemic?
Will we just have to get used to the fact that there’s always a few meme stonks that are gyrating wildly up & down? A few flare ups here and there?
Has an equilibrium been reached in this new world, or is the current state unstable, and will we see either a fall of WSB’s influence, or a rise of it as it pulls ever bigger stunts?
Elastic Q1 Highlights (Fiscal Q4, whatever…)
(I say it again: Let’s kill offset fiscal years, and all use calendar years)
First, let’s do a little victory dance on the easy call I made last Q when they guided to 28% growth:
They beat that guidance by 16% on the top line…
Some stuff from the call that I found interesting:
SaaS revenue was $51.3 million, an increase of 77% year-over-year
As of Q4, our Elastic Cloud business now has an annual run rate of over $200 million.
So this smaller part of the business is growing about twice as fast as the rest… How long until it becomes a much more significant part, and most importantly, is its growth rate decline slope more gradual than the rest of the business?
That’s going to matter…
Security is our fastest-growing segment within our business. Observability represents north of 40%; Enterprise Search, less than 33%; and Security represents around 20% of our business now, and it's our fastest growing. But we treat all of that as the same data. [...]
Last quarter, we launched this sort of ransomware protection, for example, on the endpoint itself and bringing similar protections to the endpoint to the data center and to the cloud.
They’re not as well positioned to benefit from the security investment tsunami that is coming as companies like Microsoft, Crowdstrike, Zscaler, Okta, Cloudflare, etc, but if you take into account the fact that they barely had a security offering not that long ago and that it’s already a fifth of their business and growing faster than the rest, I’d say they’re doing pretty well.
And they’re moving fast and can improve their positioning faster than many of the legacy competitors competitors that are losing share to the ones I mentioned above.
The rising tide should lift Elastic’s security boat too. 🚤
Back to the ☁️ :
where can we provide the best service for our customers, everything else being equal, we definitely believe that our cloud service provides the best service for our customers. We are the creators, maintainers and builders of our products, and we know how to manage them well, and we know how to build cloud dedicated features that help, for example, scale them automatically, upgrade them and manage them efficiently at almost like a worry-free experience to our user base. [...]
a big part about our cloud strategy is to be there for our customers wherever they are. That's why we're partnering with all cloud providers [and offer on-prem] [...]
And it’s not just one side cannibalizing the other:
the cloud strength that you saw was not from any transitions in the customer base. A lot of that was net new adds. Of course, we also saw strength on the self-managed side of the business
If they can keep differentiating from AWS’ fork of their code-base, and improve at a higher velocity than them so that the delta between the two keeps increasing, it feels like over time most of their customers will be on their cloud, and that AWS will lose the elastic wars eventually and may decide that it’s better to just do what Azure has done and partner directly with Elastic (it’s not like that wouldn’t have plenty good margins for AWS too).
Salesforce’s 🐊 Mulesoft Acquisition
Jamin Ball of Altimeter:
Salesforce's acquisition of Mulesoft for $6.5B 3 years ago (May '18) was such a massive steal, and great strategic move. In Q1 Salesforce announced Mulesoft contributed $380M in rev, growing 49% YoY. What would that business be worth today? $40B? $50B?
For reference, Okta did $250M of rev in Q1 growing 37% and is worth ~$34B
Datadog did $200M of rev in Q1 growing 51% YoY and is worth $28B
At a similar multiple to Okta and Datadog, Mulesoft would be worth $50B today
Of course, there’s likely plenty of revenue synergies that came from plugging into Salesforce’s distribution engine, but as Jamin says: “the synergies are what make it such a perfect acquisition!”
Science & Technology
I thought this was interesting…
This has many advantages, but also incurs speed penalties (it takes more CPU cycles to parse and interpret it, so it runs slower than a compiled binary blob, which is what most apps are in their final, executable form).
V8 does a lot of fancy stuff to determine what falls into bucket #1 or bucket #2 to try to make things as fast as possible.
But now, they’ve created a third bucket, with a new step in the compiler pipeline caller Sparkplug.
It’s a “non-optimizing” compiler, which sounds weird until you understand the trade-offs that it makes:
With our current two-compiler model, we can’t tier up to optimised code much faster; we can (and are) working on making the optimisation faster, but at some point you can only get faster by removing optimisation passes, which reduces peak performance. Even worse, we can’t really start optimising earlier, because we won’t have stable object shape feedback yet.
Sparkplug is designed to compile fast. Very fast. So fast, that we can pretty much compile whenever we want, allowing us to tier up to Sparkplug code much more aggressively than we can to TurboFan code.
So the idea is basically to be able to compile code into a format that is still faster to execute for a CPU than non-interpreted code, but that is much faster to compile than the fully optimized code of the fully-featured compiler.
Basically, it’s like product segmentation, but for code optimization. Very cool stuff to squeeze out as much performance as possible out of the engine.
Based on the V8 team’s benchmarks (which they try to make as real-world as possible), improvements range from 5-15% (that’s big for this kind of compiler stuff), which is awesome.
And what’s great is that V8 is open-source, like the core Chromium browser, so these benefits are going to end up in all kinds of places, like Microsoft Edge (my main browser — based on Chromium, nothing to do with old Microsoft browsers, I hated IE as much as anyone…) or Cloudflare’s Workers edge computing platform, which uses V8.
‘Why are modern 50mm lenses so damned complicated?’
In the last decade, 50mm lenses have gotten more complex. [...] Three major manufacturers have recently introduced even more complex-design 50mm F1.2 lenses, containing 13 to 18 elements with multiple aspheric and low dispersion elements.
Lens designers didn't add all that extra glass just to charge more money and make the diagram look cool. These newer, more expensive, and more complex designs are supposed to overcome some of the limitations that we used to see in ultra-wide aperture, 50mm primes.
Fake Interview -> ID Theft
I enjoy learning about scams and cons, partly because they often are clever and interesting in themselves, but also because being aware of the general principles that underly most of them helps avoid them.
Here’s one that I had never learned about before that is apparently getting more popular:
One of the oldest scams around — the fake job interview that seeks only to harvest your personal and financial data — is on the rise [...]
“The endgame was to offer a job based on successful completion of background check which obviously requires entering personal information,” Gwin said. “Almost 100 people applied. I feel horrible about this. These people were really excited about this ‘opportunity’.”
How do you spot this, other than just general spidey sense? FBI has some advice:
-Interviews are not conducted in-person or through a secure video call.
-Potential employers contact victims through non-company email domains and teleconference applications.
-Potential employers require employees to purchase start-up equipment from the company.
-Potential employers require employees to pay upfront for background investigations or screenings.
-Potential employers request credit card information.
-Potential employers send an employment contract to physically sign asking for PII.
-Job postings appear on job boards, but not on the companies’ websites.
-Recruiters or managers do not have profiles on the job board, or the profiles do not seem to fit their roles.
MDMA vs PTSD
The most recent study of MDMA-assisted therapy for PTSD coming out of the Multidisciplinary Association For Psychedelic Studies (MAPS) demonstrated safety and significantly improved participant PTSD severity and functional impairment compared to the placebo with therapy treatment. The positive results of this study means that MDMA-assisted therapy treatment is halfway to FDA-approval for therapeutic use for PTSD, which requires two positive phase 3 trials. There is currently a second phase 3 trial underway in the recruitment stage and is expected to finish in 2023. For the time being, the alternative treatment remains an FDA-designated breakthrough therapy, which means it is on a fast track for experimental research and approval.
Unlike psilocybin, which we discussed last week, MDMA is not psychoactive—that is, it does not produce any hallucinations. However, it does produce a very noticeable increase in empathy, hence its descriptor as an “empathogen,” which is quite unmistakable to most. It’s this profound increase in empathy that almost assuredly plays a role in the efficacy of MDMA for aiding in the repair of trauma. To give you a sense of how MDMA is used clinically, and what the patients experience, this short (9 minute) video is a great overview. (Source)
The Arts & History
‘The End of the Tour’ (2015)
I ended up re-watching the film about a few days in the life of David Foster Wallace and Rolling Stone writer David Lipsky on Wednesday.
It was very enjoyable. I suppose that if you go in with the wrong expectations, it could be a disappointment, so if you’re giving it a try, know that it’s basically just two dudes spending time together and talking about life, writing, the human condition, how the modern world can affect our psyche, and dancing around feelings of jealously/insecurity/etc.
It’s not the best film ever, but it’s a very honest film, and I liked living through that slice of life.
I also like that it shows characters and environments that we don’t often see in films. DFW’s house is glorious in its real ordinariness (unlike what Hollywood usually shows us as a typical house), the small moments when they’re eating, smoking, ordering, talking to strangers… It’s all in the little things.