Patrick McKenzie Profile picture
I work for the Internet, at @stripe, mostly on accelerating startups.
Rajesh Kuppuswamy Profile picture Jeremy Pinnix ☧ Profile picture Michael Hood Profile picture 🇺🇦 Profile picture John-Kim Murphy Profile picture 120 added to My Authors
25 Nov
I hate that this is true, but instrumentally useful: if you ever experience the symptom “I tried to get in contact with a firm but no human will talk to me” your best bet is stop following the obvious process and start cold emailing people arbitrarily high up ladder (tech) or…
… groups which by their nature need to have an open inbox and who are failing if they optimize for resolution speed over resolution accuracy (Legal, Compliance, Investor Relations, etc).
This is a) downstream of a business decision to fob certain customers off at scale and b) functions as a class competence check because firm institutionally assumes that if you’re important enough to talk to you know this w/o needing it explained to you.
Read 6 tweets
25 Nov
In central Tokyo unexpectedly (the pandemic has cut down on seeing the city, sadly) and was struck by two beautiful things so I thought I’d share:
I passed the HQ of my friendly neighborhood logistics firm. It is in very, very expensive real estate.

The first floor has a public accessible customer service counter in case you want to e.g. give them a package.
This is operationally inefficient, but I love it as a statement piece:

All the execs, salesmen, accountants, etc pass drivers and customers on the way up to work every day, and they’re reminded that nothing elsewhere in the building is more important than accepting packages.
Read 7 tweets
23 Nov
The piece about the startup working environment currently making the rounds includes several anecdotes of good startup management, e.g. clearly communicating expectations and management intervention with employees who were clearly pushing themselves in an unsustainable fashion.
I’m not linking to it because it’s a non-story. You get all the signal from the phrase “Startups can sometimes be intense work environments. Here’s a startup.”
In the middle there are some anecdotes to support the indictment the author is attempting to make of the culture, e.g. someone was working during their wife’s labor until senior execs stepped in and told them to please log off.
Read 4 tweets
19 Nov
We're approaching tax season for regular employees in Japan and I'm updating my Slack dingwords to include all of my weird but suddenly acutely relevant hobbies.
"Ah yes you would think that iDECO is 個人年金 since it is, after all, a pension (年金)which you pay for for yourself(個人), but you'll copy that information into the 小規模企業共済 box instead. Why is a fascinating story; do you want to hear it?"
It's probably not my most leveraged way to make a dent in geeks' financial futures but the salaryman in me does think that I owe junior coworkers literally up to the legal limit of Not Investment Advice so they can scoop up the sweet sweet ~$400 of tax benefit available.
Read 4 tweets
18 Nov
I feel unreasonably proud as a Chicagoan interested in credit cards that when I got the email about Reserved by Sapphire I said “Chase did Tock? No. Chase must be white labeling Tock.”

(They are.)…
Tock is an amazingly brilliant business by the way, by the sort of restauranteur who also posts on HN for fun.

High end restaurants have an inventory management problem and the relevant inventory is seatings not food. Tock suggests they sell pre-paid reservations like shows do.
This way they don’t have to eat no-shows or parties which reserve a table for 8 but only show up (and pay for) 5.
Read 5 tweets
17 Nov
Stripe Tax is chugging along. We wrote a brief roadmap of what problems we can solve for users today and which ones we hope to be able to solve soon:…
Some people think that taxes are complex because the people who write or enforce the rules are perverse. I think it’s mostly because a) taxes are used to express society’s preferences and b) society’s preferences are complex.
For example, consider sales/use taxes around food. A key goal of most tax regimes is “progressive taxation”; you generally want to take more as a percentage from “richer” people and less from less well-off people. (Definition of “richer” also complex when you try to encode it.)
Read 17 tweets
5 Nov
If you enjoy my writing and enjoy e.g. Money Stuff you would likely enjoy Net Interest, which is @MarcRuby's newsletter on banking topics.

Let me illustrate with a single paragraph why this is obviously the future of e.g. equities research:
Substack has made it incentive compatible for the world's various obsessed experts on various topics to hire themselves out to the Internet rather than hiring themselves out to e.g. financial intermediaries, and then they do the sort of deep work that experts do.
Also: holy me do I not want to be the investment banker who let a math error work its way into an IPO prospectus.
Read 6 tweets
5 Nov
Agree so much. Interesting knock-on effects, too:

1) Companies which previously had a $25k minimum to account for legal friction/founder headaches should be happy to take $1k from e.g. an engineer.

2) It's SO MUCH better for "not my day job" angel investors to centralize info.
e.g. I owe the National Tax Agency (Japanese IRS) some math every year about all of my overseas assets, including those which did nothing in a year, and the process of collecting that information looks something like this:
Non-AngelList investment from 2012: get docs out of Dropbox, grep inbox to see if e.g. this is the one that I remember signing the docs about a corporate transaction this year or not, math math math, en-spreadsheet.

AngelList: download report on all, send to accountant, done.
Read 11 tweets
5 Nov
Bits about Money #3: How credit cards make money.…

Some brief elaborations from the cutting room floor:
The essay talks about cross-subsidization at a few points. One fascinating form of cross-subsidization was that credit cards *changed who ultimately pays for an individual's use of credit.*
This had huge ramifications for small businesses, which are historically (and currently) horrifically capital constrained almost all of the time. They also have extreme difficulty in lining up traditional bank financing.

Credit cards let merchants opt-in to financing customers.
Read 11 tweets
5 Nov
A quirky feeling I've had on our covid-19 response is that some institutions have difficulty doing things which are *clearly* within our capabilities while others are pushing boundaries in their respective fields.

A happy example of the later:…
More effective covid-19 treatments is always good news, of course, but taking a teensy look at the actual specifics of this:

This was identification of a new molecule, synthesization, testing, demonstrating of efficacy, and preparation for scaled manufacturing.

In two years.
My impression is "Humanity very literally did not know it could do that on that timescale. Not for covid. Not for anything."
Read 4 tweets
4 Nov
Isn't this just *obviously* the way the typical consumer's most important transaction should want to work? People are worried if they can make the math work. Me, less so, but that aside: few would say "Sign me up for the traditional sell/buy process!" with this as an option.
Ignoring the "new experience" thing which is product speak for "We stitched everything together in a web app", substantively:

1) You ask Opendoor for a quote.
2) They give you a hard quote and accept sale contingent on you winning target house.
3) You share pre-qual letter.
A "pre-qualification" letter is a document from a mortgage broker or originator that says "Contingent on you submitting a bunch of documentation, indicatively, we think we can underwrite you for a mortgage up to $X." Most common use is showing to seller to say "I could swing it."
Read 12 tweets
4 Nov
*picking Lillian up from school*

Construction worker: Sensei, is that the last of them?
Me: Pardon?
CW: The kids. Are most of them [past the construction site or should I stay directing traffic]?
Me: I'm afraid I don't know; I came from the other way.
CW: Not from school?!
Me: No.
CW: You're not the English teacher?
Me: No, she's a young Filipina woman and I *gestures*.
CW: ... So you're not a teacher?
Me: Correct.
CW: Why the PTA badge then?
Me: ...
CW: OH I GOT IT. Sorry. Thanks.
And *sigh* the possibility of this dialog happening with a police officer is why I am very, very careful to put on my PTA badge prior to getting close to the school.
Read 5 tweets
4 Nov
I'm doing some podcast editing (for the first time in almost five years!) and Descript ( ) is as close to magic as anything I've ever used.

There's some AI/ML under the hood which lets you edit audio as text. It's mindblowing.
Auto-generated transcript includes:

"Uh that's an interesting question, let me think, OK, [actual content]"

Highlight first part of answer, hit delete; it makes a seamless edit between the question and the contentful part of the response.
Not quite as good as an NPR editor but muuuuuuuuuuch cheaper and faster, and you really have to be familiar with Descript and listening closely to know it happened. (I periodically smile when I "catch" another podcast I listen to clearly using it.)
Read 6 tweets
3 Nov
I agree with much of this thread, and think it is an opportunity for folks to write about upleveling engineering practices at smaller orgs.

That is not the way that incentive gradients run for engineers at AppAmaGooFaceSoft, to be true.
But there’s a huge amount of value in e.g. teaching Rails shops enough about Ansible and Vagrant so that they’re able to adopt the “cattle not pets” treatment of servers, dev boxes, and laptops that is standard at companies with teams working on developer productivity.
And as someone who will certainly have a shrine to Hashicorp built in the corner of any company I ever run in the future, I do think it actually is possible to have a fair degree of rigor and engineering skill at companies with 1, 2, or 12 engineers.
Read 6 tweets
2 Nov
Tailscale, Magic Wormhole, and ufw (Ubuntu's I-can't-believe-its-not-iptables) finally make me feel like I have something approaching rigor in my personal and project networking, and are *so* much easier than my previous methods for achieving this.
I wouldn't recommend them for consumers but if you are e.g. running a business and want the thing a networking team would ensure you have w/o having a networking team, these are *amazing.*
Tailscale makes setting up VPNs trivially easy; this lets you keep almost all of your infrastructure strictly off of the public Internet.

ufw just decreases pain of software firewall administration.

Magic Wormhole makes getting things between VPNs for e.g. operators trivial.
Read 5 tweets
2 Nov
The Nubank F-1 (registration statement similar to S-1) is extremely, extremely interesting if you geek out about either neobanks or the possibilities of fintech around the edges of the legacy financial system.…
The highlighted sentence is, ahem, extremely favorable. Image
The sentence immediately after it is not, and since this is commonly misunderstood by startup founders it is worth mentioning this:

If 90% of your customers are coming to you organically and you have a successful paid marketing channel *one of those two claims is not true.*
Read 4 tweets
30 Oct
I broadly endorse this thread as descriptively accurate.

Some of the many implications of filtering by pedigree:
1) Not a new suggestion but *do not optimize job searches around unsolicited resumes.* Find someone inside target firms, and almost literally anyone in engineering is better than /dev/null, to get excited about you working there.

2) Given “plausible” is essentially binary…
… people early in their career may want to intentionally spend first 2-3 years on getting one plausible name versus whatever your top option would otherwise have been.

3) If you are hiring for a startup, resumes of good engineers who’d not read as plausible are your opportunity
Read 10 tweets
29 Oct
As someone who was recently on the other side of the tech money vs. charity table, one interesting learning was how accessible Donor Advised Funds are (for donors) and how ridiculously preferable they are to charities versus chasing individual checks.
I’d probably advise anyone to have one if you consistently give away more than $1k or so a year, particularly if that is to multiple recipients.

The commanding favorite among donor base was Fidelity Charitable, which (personal opinion) as a CEO I felt was wonderful to work with.
“What do you mean ‘work with them’?”

The first time you get a donation from any donor using a DAF they teach out and ask for your IRS paperwork and some relatively basic KYC and (optionally) banking information. Lead time of a few days.

2nd and subsequent: money shows up.
Read 4 tweets
29 Oct
Following up on the C.R.E.A.M exploit looks like it might be an interesting news day in cryptoland.
Not trying to chortle here; the choice to make all transactions public in real time is an interesting one and does have novel implications for e.g. responses to crises, ability to coordinate during a stressful event, contagion in a panic, etc.
You could imagine an alternate universe where the AP published “Breaking: Chase has called all deposits and loans custodied at Bank of America, causing a spike in BAC credit default swap rates.”
Read 5 tweets
29 Oct
I am optimistic about the pace of innovation in financial services, and importantly, much of what we've wanted for decades *is already happening.*

This is often underappreciated, so I covered it for the second issue of Bits about Money.…
One of the reasons this is undercovered, by the way, is that there is a huge difference between the socioeconomic position of the financial commentariat the challenges (and solutions) faced by e.g. Americans at the 20th percentile in income.
Sometimes this causes people to underappreciate how difficult life is when one doesn't have a job in tech. That factor is well-appreciated. Less well appreciated is that it causes people to underappreciate how towering an achievement e.g. Cash App is.
Read 4 tweets
29 Oct
Also, and forgive me because this sounds obvious but apparently it is not:

You should have internal tools, or a sub team, own the integration of these with internal tools, and those integrations should be as good as your shipping products are.
“What’s your favorite example at Stripe?”

There is a heuristic that runs “Is this necessarily going to be a Zoom meeting?” and updates calendar invites with Zoom details, and also a button with instant make-it-Zoom (not sure if that is via extension or otherwise).
In the Before Times this was a godsend to us remote workers because it was always socially awkward for the 1 remote in an 11 person meeting to say “Hey folks could I pretty please get you to turn on the videoconferencing hardware” and thanks to that work it would be default done.
Read 4 tweets