Like most people, I get too much email. Over the years, I’ve tried several approaches to clean up and manage the long list of messages that accumulate in my inbox. I’ve defined inbox rules (“Filters in Gmail”), reported individual emails as spam, and clicked “Unsubscribe” on promotional emails. Yet, new unwanted emails keep arriving, which require more tedious management.

SaneBox and the Pitfalls of Machine Learning

I learned about SaneBox and gave that a try. At first, it seemed to help. It removed many non-essential emails from my inbox putting them in special folders like “SaneLater” and “SaneNews”. But I found it also required a lot of maintenance and I was often finding unwanted emails still arriving in my inbox, and worse, important emails would sometimes get moved to SaneLater or SaneNews and I’d miss them for a week or more.

SaneBox uses machine learning, among other techniques, to automatically categorize and label email messages. You can help train it by overriding its decisions - removing a SaneLater label or moving a message to SaneBlackHole. But I found that ineffective. I had a reminder each week to review all the messages in SaneLater and SaneNews and either leave them, remove the label, or move them to SaneBlackHole. It was a lot of tedious work and it seemed to never really get that much smarter about its rules. Week after week I found myself repeating the same process on emails from the same source. It wasn’t helping me. I was still doing all the tedious work to review and manage emails, but the work was now happening across three folders instead of just my Inbox.

All machine-learning solutions suffer from a tradeoff between false positives (legitimate emails marked as spam/non-important) and false negatives (spam/non-important emails not marked as such). If you tune the algorithm to try and lower one type of error, you increase the rate of the other type of error. And as a feature in an app, you have to live with the tuning that the developer choose and deal with the resulting errors as they arise.

I wanted a solution that was more explicit, easier to understand how it was applying rules, and didn’t require as much oversight and maintenance. After searching around and researching different applications that address this problem, I came across Mailman.

Mailman - Simple and Effective

Mailman takes a different approach to this problem using a simple workflow that covers four cases:¹

1. Important emails pass through directly to your inbox. These are called “VIPs” and use explicit rules for domains, senders, or keywords.
2. Emails that you have explicitly blocked are labeled as “Blocked by Mailman” and placed in a “Blocked by Mailman” folder.
3. Emails that you have explicitly allowed are labeled as “Mailman” and held to be delivered to your inbox on the next delivery schedule.
4. Any new emails that do not match any of the other three rules are held for review once a day in the Daily Digest.

This is primarily a whitelisting approach to emails - only people that you explicitly allow are placed immediately in your inbox. But it adds a “hold-and-deliver” feature for other emails you want to receive, but which don’t require your immediate attention. These two rules work together that to your inbox uncluttered. I rarely have more than a handful of emails in my inbox at any time now. I “process” them soon after they arrive (respond if needed, add to my to-do list and archive if I have to take action, or archive if they require no action).

The other nice feature of Mailman is the Daily Digest workflow. It presents all your new messages in a web app that you can quickly go through and allow or block. You can also do this in bulk to speed things up the first few days when you tend to have more messages in the digest. In contrast, SaneBox had no tool to “triage” new emails each day. It was left as a workflow for you to define on your own, jumping between your email client and sometimes back to the SaneBox web app to tweak rules. I found that added friction resulted in me doing the maintenance work irregularly, which led to me falling behind even more.

Back in Control

I’ve been using Mailman for about a month now, and it’s been working extremely well for me. No more missed emails that caused me embarrassment — or worse — made me look lazy or unprofessional. No more repetitive, tedious cleanup of mail messages, and when I see a new email in my inbox now, it’s something I want to see.

Inbox Zero used to be an aspiration to me — a nice idea but always out of reach. Now it’s my normal state and I no longer have to think about it.