Monthly Archives: September 2020

Friday Updates: Data Retention at Preceden, Closing Down Timeglider, Organizing Looker, the Psychology of Money, and the 3m Mario Speed Run

What I’m Working On

At Preceden, I’ve been thinking a lot about data retention. For example, it has users that signed up a decade ago, used it briefly, and then never again. Unless they deleted their account, their user and timeline data still exist in Preceden’s database. In the past, I never gave any thought to whether it made sense to keep this data around forever. Maybe the user eventually returns one day, so why not keep it around? Analyzing all of that data has also been tremendously valuable. And for publicly shared timelines, there are SEO and user acquisition advantages to keeping those timelines around. But imagine a person that signed up in 2012 to create a private timeline about a messy divorce and hasn’t used it since then. Should Preceden retain that data forever? Probably not.

Addressing this is way easier said than done though. After I tweeted about this dilemma, Emilie Schario pointed me to an insightful Basecamp podcast episode about how they built something called the Data Incinerator to deal with this at their company. It’s worth a listen for anyone thinking about data retention. I’m probably going to build something like this to improve data retention in Preceden in the coming months.

—

Last summer I acquired Timeglider, a competitor timeline maker tool. I recouped the acquisition cost in two ways: I kept Timeglider running for its existing customers (which have continued to bring in revenue) and also redirected Timeglider’s homepage to Preceden (which brought in a lot of new customers for Preceden).

Because Timeglider hasn’t had any new sign ups in the last 14 months, its recurring revenue has gradually dwindled as its customers have churned. So the question I’ve been debating for a while now is how long to keep Timeglider running. On one hand, maintenance and costs are minimal, so it’s basically a small stream of completely passive income that would probably continue for years. On the other hand, the remaining recurring revenue wasn’t significant, most of the customers were inactive, its codebase and infrastructure are very brittle, and operating it takes up headspace that I’ve love to free up. After weighing it all, I decided earlier this week to announce it was shutting down at the end of November.

Executing on this plan took more work than I anticipated though.

  • Many Timeglider customers paid annually. If someone paid $50 for access to Timeglider through August 2021, but I shut down the service in December 2020, it would be unethical to keep their full payment. I wound up issuing a lot of prorated refunds for recent annual payments.
  • Stripe doesn’t provide a way in the UI to cancel multiple subscriptions at once. The quantity was more than I wanted to do manually, so had to write a little Ruby script to iterate over all of the subscriptions and cancel them.
  • I added a prominent notice at the top of Timeglider with a link to FAQs about the closure including instructions on how to move their Timeglider timelines to Preceden.
  • I had to notify all of the customers whose subscriptions I cancelled to make sure they knew the site was shutting down. Timeglider also has a free plan which had some active users, so I had to email them as well. Probably not the most efficient way to do this, but I BCC’d all of the impacted users 25 at a time in Help Scout (its BCC limit), using the Saved Reply functionality to compose each message.
  • Most of the responses from users were understanding and they appreciated that I was giving them notice and also provided a way to move their timelines over to Preceden.
  • There were some bugs I had to deal with though. By cancelling everyone’s subscriptions, it reverted everyone to Timeglider’s free plan which had feature limits including the inability to access more than 3 of their timelines. For users with more, this meant they couldn’t export their data. Had to hardcode in some logic to give everyone access to Timeglider’s top premium plan until the site shuts down.

Running that Ruby script to cancel all of the subscriptions was a little bit painful, but in the end I think shutting it down will be a good decision.

—

In my ongoing machine learning journey, I finished DataCamp’s Feature Engineering for NLP in Python course and started on its Introduction to TensorFlow in Python course. 8 courses including this one to completely finish the ML track.

—

At Help Scout, I started on a project to clean up our Looker instance. We’ve been using Looker for business intelligence reporting for more than 3 years and over that time we’ve accumulated hundreds of dashboards and more than a thousand individual Looks. We mostly try to organize these by function: a folder for marketing reporting, a folder for sales reporting, a folder for finance reporting, etc. But over time it’s gotten very messy. Lots of unused dashboards and Looks. Lots of duplicate reporting. Even some completely empty folders. For any employee venturing into Looker to try to understand our metrics, figuring out where to go was a challenge to say the least. I’ve been cleaning it up and with just a few hours of work so far it’s way, way better than it was. I regret not spending more time on it earlier.

What I’m Reading

The Psychology of Money by Morgan Housel. Really enjoyed this book. Lots of sage advice on how to think about personal finances, uncertainty, risk, etc. Morgan is worth following on Twitter too. And for you podcast fans, here’s an interview with him from earlier this month (thanks Jason for the recommendation).

What I’m Watching

This deep dive explanation of a 3 minute Super Mario Bros 3 speed run is mind blowing:

I thought it was just going to be someone playing through the game perfectly, but it’s so much more than that. The record is made possible by painstaking work to reverse engineering the game to figure out and exploit a very subtle glitch. Here’s the HackerNews discussion about it.

What Else

I wrapped up the 28-Day Keto Challenge. Writeup here. After being back on my normal carb-heavy diet now for 5 days and feeling like a glutton, I’m giving serious thought to going back on keto. We’ll see.

Hope everyone is doing well. Thanks for reading.

28 Days of Keto with the Wearable Challenge

In my ongoing effort to make the most of quarantine, I just wrapped up a 28-day experiment trying a keto diet. This diet had a twist though: if I stayed below a certain blood glucose level every day, I stood to make $700. Kind of.

Wearable Challenge

Freestyle Libre is a Continuous Glucose Monitor (CGM) which means its a sensor you wear to continually monitor your blood glucose levels.

The Freestyle Libre CGM sensor. It’s painless to apply and you barely notice it.

It’s intended for use by type I or type II diabetics, but this startup called Levels is repurposing it for non-diabetics to help them improve their diet and health.

And last but not least there’s the Wearable Challenge, a group that has partnered with Levels to help people adopt Levels to improve their health.

Here’s how the Wearable Challenge works:

  • You pay the Wearable Challenge team $850 after you’re accepted into the program.
  • They work with Levels to mail you two 14-day CGM sensors.
  • Every day for 28 days, you have to wear a sensor and scan it periodically with an app on your phone.
  • For every day that you stay below your max target blood sugar level (120 mg/dL) and meet certain other requirements (taking a picture of everything you eat, not having any gaps in your data) you get $25 back. If you do this every day for the entire challenge, you get a full $25/day * 28 days = $700 back. The remaining $150 pays for the sensor and you don’t get it back.

I’m not diabetic nor do I have a ton of weight to lose, but I’ve wanted to experiment with keto for some time and the Wearable Challenge seemed like an innovative way to go for it. I applied to the program, was accepted, and officially started the challenge on August 17th.

Keto

On a keto diet, you drastically reduce your carbs in your diet so your body switches over to using fat as its primary energy source.

I don’t enjoy cooking so headed to YouTube while preparing for keto to plan for what I’d eat throughout the challenge. I found this meal plan by Water Jug Fitness which I wound up following pretty closely throughout the challenge:

My meals look like this:

  • Breakfast: Omelette with two-eggs, mushrooms, spinach, and cheese with a side of cooked spinach, avocado, and two pieces of bacon.
  • Lunch: Taco salad with ground beef, onions, romaine lettuce, baby tomatoes, sour cream, avocado, and cheese.
  • Dinner: Some combination of a protein (fish, steak, chicken) with a side of vegetables.
  • Snacks:
    • Almond milk protein shake with MCT oil
    • Almonds
    • Pork rinds 😋

I usually meal prepped for lunch. I’d prepare 4 days worth of taco salad ingredients (by cooking the ground beef, cutting up the lettuce, etc) and placing them in a meal prep storage container which I’d put in the fridge and grab for lunch each day. This was really helpful because it made it super easy to stick to keto when normally I’d be drawn towards some less healthy meal for lunch.

Taco salad for lunch every day

While it may not work for everyone, I was mostly fine with having the same meals for breakfast and lunch every day with a little variety for dinner.

Scanning the sensor

The Freestyle Libre sensor can only hold 8 hours of blood sugar readings, which means that you have to scan it with your phone at least that often otherwise you’ll have gaps in the data (the Wearable Challenge team did make exception for gaps overnight though since depending on how long ou slept you might have some gaps in the data right after you headed to bed).

Every few hours I would load the Freestyle Libre app on my phone and hold it up to the sensor on my arm. It would scan the sensor and show me my blood sugar level along with a chart showing the trends that day:

For most of the challenge I stayed well below 100 mg/dL, but toward the end I started incorporating more carbs which had my numbers jumping to 110 mg/dL here and there.

The Freestyle Libre app is set up to pass data along to Levels, which also has an app:

It’s the Levels app that you use to take photos of what you’re eating, one of the requirements to get the $25 for that day.

The Wearable Challenge has access to the data in Levels so they can monitor how you’re doing to determine whether to pay you $25 for that day or not. Even though the Wearble Challenge is fairly new (I was part of the third cohort) I’m assuming most of the day to day management of it are automated at this point.

Keto Impact

The first week of switching to keto you go through something known as the keto flu which is a result of your body adapting to the new diet. I had many of the symptoms: dizziness, keto breath, sugar cravings, and headaches. After the first week though most of it diminished significantly.

One thing that really blew my mind was how little I thought about food after the first week. I suspect that’s part of keto in general, but also because I was eating the same meals almost every day. Food just got very… boring. Cravings and hunger almost completely disappeared.

I didn’t exercise at all during the first week, but after my body adapted to keto I picked it up again, doing 20-30m YouTube HIIT workouts a few times per week. I expected going into this that working out would be tough due to to low energy, but wound up not having any problems with it.

I actually had a lot of energy despite only consuming 40-50g of carbs per day. Your body really does adapt.

Most of the last ten years I’ve been around 185 pounds, varying a few pounds in either direction. At the start of the challenge I was on the low end at 183.2 pounds. After about 5 days, I was down to 179 pounds, and then slowly got down to 178 by the end, the lowest I’ve been since college.

What’s Next

In the end, I managed meet the requirements each day and will be getting the full $700 back:

But I’m not going to stay on keto. I enjoy too many non-keto foods and am healthy enough where it’s not a huge deal to have some excess sugar in my diet right now. That said, I hope to find some balance between my previous carb-heavy diet and keto.

And who knows, if quarantine goes on long enough, maybe I wind up going back on full keto and trying to get down another few pounds. We will see.

Kudos

This little adventure wouldn’t have been possible without the Freestyle Libre, Levels, or the team organizing the Wearable Challenge. Huge thanks to everyone who helped make it happen. Also, shout out to my wife for putting up with my keto breath for most of the last month 🥰.

If the Wearable Challenge sounds intriguring to you, I’d encourage you to check out the website and consider applying.

Friday Updates / 2020-09-11

What I’m working on

At Help Scout, a customer using Beacon wrote in asking about a warning that Chrome was displaying on their site:

The root cause seemed to be some attribution tracking cookies that were set by a script I worked on, so the support request made its way to me. Digging into it, we can see a lot of Help Scout cookies (including attribution tracking ones like _A_FirstTouchURL) are being set when Beacon is loaded on the customer’s site:

Despite having worked a lot with cookies and tracking scripts in the past, I was fuzzy on how these cookies were being set. Why, for example, were there Mixpanel and Google Analytics cookies when Beacon wasn’t loading them? Turns out that because the customer is loading Beacon from a helpscout.net domain, all of the helpscout.net cookies are made available on their site unless they’re explicitly set the cookie not to be passed along.

This can be achieved by setting the SameSite cookie attribute to Strict. In this case, there’s no need for most of these helpscout.net cookies to be passed along to third party sites loading Beacon, so we’re going through the steps to mark the cookies as strict when possible. Some cookies on helpscout.net are set by third party scripts and can’t be set to strict, so some amount of cookies getting passed along is inevitable, but at least we can minimize it.

This week I knocked out two DataCamp courses in my machine learning adventures: Hyperparameter Tuning in Python and Introduction to Natural Language Processing in Python. Typically I only am able to finish one course per week, but a lot of the material in these was covered by other courses so it made these courses fairly quick to work through.

One new thing was the introduction of TPOT, a tool that uses genetic programming to find an optimal classifier and hyperparameters for a given data set.

Long time readers may remember my Evolution of Color project in 2014. It uses genetic algorithms to evolve a population of colors towards a goal color. Unlike most of the other Emergent Mind projects that were I re-implemented other people’s projects in JavaScript, the Evolution of Color was completely original, making it one of the projects there I’m most proud of.

I hope as I continue down this machine learning path I get more opportunities to work with genetic algorithms.

And on that note, I’ve been mulling over where to focus in the coming months to keep leveling up my machine learning skills. Currently my plan is:

  • Finish the 11 remaining machine learning DataCamp courses by end of the year
  • Spend a few months in early 2021 focusing on Kaggle competitions to gain experience applying what I’ve learned
  • Figure out how to integrate machine learning features into a web application, whether it be for Preceden, a new app, or maybe even Help Scout if there’s an opportunity. Apparently deploying ML applications is quite difficult though. We will see.

What I’m watching

My wife and I just finished watching Umbrella Academy on Netflix:

It’s about a dysfunctional family of superheroes that need to work together to save the world. I enjoyed it and would recommend.

Product Recommendations

I’m in a mastermind group with Tom Davies and Jason Rudolph. Tom has two popular Shopify apps, Best Sellers and Flair, and Jason has BuildPulse, a tool that helps automatically identify flaky tests (automated tests that randomly pass sometimes and fail other times).

If you run a Shopify store or your team is banging their head against the wall dealing with flaky tests, their tools are definitely worth checking out.

What else

I’m on day 26 of a 28-day keto challenge. I’m participating through the Wearable Challenge initiative which basically means I’ve been wearing a Continuous Glucose Monitor (CGM) on my arm and for every day that I stay below a certain blood glucose limit I get $25 on my own money back. I’ve lost about 5 pounds and feel pretty healthy, but probably won’t continue keto after the challenge ends in a few days. I’ll write more about this whole thing in a separate post.

Hope everyone’s doing well 👋.

Friday Updates / 2020-09-04

I’m going to try a little experiment for a few weeks and post a short update every Friday about what I’m up to.

I enjoy writing and used to blog here a lot, but family, work, and projects tend to take up a lot of my time these days so I haven’t prioritized blogging for a long time. These short updates are a way to get me back into writing and also serve as a way to stay closer with you all. I’m time-boxing myself to about an hour every Friday to write these so we’ll see how they turn out :).

What I’m working on

At Help Scout, I’ve spent a lot of time recently helping with a big Go To Market (GTM) stategy project. A GTM strategy is a comprehensive plan for launching a new business or growing an existing one. Help Scout has been around for more than 9 years so we have a lot of data that we can use to understand what industries our product resonates in, how features are being used, who the buyers are, how our customers grow over time, etc. All of this can be used to help inform our GTM strategy for the coming years.

At Preceden, I addressed an issue this week stemming from a bug feature in wkhtmltopdf, a tool that converts web pages to PDF files. It’s what Preceden uses to let users export their timelines as PDFs. The problem with it though is that if a webpage contains an image that points to an invalid (404) image, the PDF conversion completely fails. This happens with Preceden because users can paste an image URL into an event’s notes and Preceden will try to display the image. But if the URL doesn’t point to a valid image, the export fails and I get support emails like this:

The link above refers to a World Civilization Preceden timeline that I want to download, but for some reason each time I try, an error message shows “There was an unknown error while processing the download. Please contact support.” I tried on multiple devices, but the same problem persists.

In the past I’ve solved this by parsing the image URLs from event notes and using the FastImage gem to verify that each one is valid. It then caches the results (valid or not valid) for each image and doesn’t attempt to render the image if it’s invalid. Problem is, I permanently cached the results of the validation check. For a timeline that’s been around for years, sometimes the previously valid image URLs become invalid. As a result, a timeline that once exported successfully might eventually start failing, leading to frustration and support tickets.

I addressed it this week by introducing some code that revalidates all of the image URLs in a timeline if an export fails.

Sometimes building a SaaS product is sexy and exciting, but often it’s fixing random issues like this.

I’m still continuing to learn machine learning, probably spending 10-15 hours/week on DataCamp, writing documentation for myself, and building small projects. This week I finished a fantastic course on Dimensionality Reduction in Python. The big idea with dimensionality reduction is that you can often reduce or simplify the inputs to a machine learning model to improve it’s performance and how well it generalizes to new data.

What I’m reading

Stumbled across Ethereum is a Dark Forest on HackerNews. I own a small amount of Ethereum but honestly have no idea how it all works behind the scenes. This article offers a glimpse of what’s going on under the hood. Eventually I’d love to dive into it more.

In that article the author mentions he enjoyed The Dark Forest, a sequl to Cixin Liu’s popular Three Body Problem novel. I read the latter a few months ago and seeing the sequal praised so highly in this post made me pick it up and start reading. Been enjoying it so far.

What else

  • My son is wrapping up his third week of virtual kindergarten today. I give the school a ton of credit, they’re really doing the best they can with it. Just sad that he’s not getting to experience kindergarten like he would in a world without Covid.
  • Speaking of virtual, I attended Microconf’s one day online conference earlier this week. I attended the live ones maybe 5 times in the past, but haven’t been for a few years. Kudos to Rob, Mike, Xander, and the rest of the team for organizing the virtual one this year.
  • If you’re trying to come up with a bilingual baby name, check out MixedName.com, a new baby name generator from a buddy of mine, Bemmu Sepponen. The service recently got a ton of attention and praise on Reddit and HackerNews.
  • And last but not least, I want to recommend Hey, the new email service from Basecamp. I was skeptical at first, but having used it for a few weeks now I’m a huge fan. The combination of their screen out feature and ability to categorize senders as Paper Trail or Feed have drastically reduced the amount of emails I’m exposed to each day, leaving me more time to focus on important things.

I hope this email finds you all well. If anyone wants to catch up sometime, I’d love to jump on a call. Drop me a note at mazur@hey.com.

Cheers!