POSTMORTEMS ★ 67 upvotes 2017

Why I shut down wizards.town and left Mastodon

Retrospective Overview & Key Insights

Distilled analysis of the business dynamics, operational challenges, and strategic takeaways:

In this postmortem from 2017, @imartin2k candidly breaks down the dissolution of their startup, examining the fatal assumptions, misallocated resources, and market dynamics that led to its shutdown.

Core Principles for Independent Builders:
1
Early signals were often rationalized away instead of prompting an immediate pivot in distribution or pricing.
2
The psychological toll of running out of capital compounds operational friction in a solo or small team setup.
3
Key Antidote: Validate paying willingness before committing months to building sophisticated product features.
Community Debate Themes:

Community discussions on HN centered on identifying product-market fit early, avoiding vanity metrics, and the emotional recovery process after shutting down.

Community Discussions & Critical Reviews

High-signal perspectives, counter-arguments, and peer founder insights debated on Hacker News (79 total discussions):

MO @moron4hire Peer Contributor

I've been a bit torn on Mastodon. It's a very interesting idea and a few performance issues aside, quite an impressive bit of technology. That federation works at all is a pleasant surprise.

I've had good conversations in the "fediverse". Engagement is super high. I've had more comments, reshares, and off-site link follows from 70 people on Mastodon than 700 people on Twitter. But I've also had a higher proportion of bad conversations than on Twitter. I guess there is a bit of a hollowing out of the middle that happens, as it takes effort to switch, so only people who feel strongly about staying on Twitter are going to go.

There's also way too much meta-discussion on Mastodon. Childish phrasing like "that bird site" instead of just saying Twitter. Too much back clapping about how amazing the fediverse is. Too much hand-wringing about what any of it means. Too much bitching about how blocking CP is "as bad as Twitter." Too much complaining about GNU Social not getting its day and trying to say Mastodon is just GNU Social (they are not the same thing, they have OStatus in common). I largely don't talk on Mastodon now because I just don't want to talk about Mastodon.

What I want right now is just my Twitter followers/followings with the 500 character post length of Mastodon. Use UX design to encourage brevity, but enable longer form when it's desired.

SI @Sir_Cmpwn Peer Contributor

This issue seems overblown, and I'm not surprised. Child porn is a great trigger word for making people approach an issue emotionally. I'd like to challenge everyone to take a few steps back and look at it more rationally instead. Many people who are concerned about their own instances probably does not have to worry [1]. I disagree with anyone taking the moral high road here too - I think it's disingenuous to call this "child porn" and illicit all of the emotional hatred that it implies when no children are being abused and it's legal to create, posess, and distribute in many countries.

The issue at hand is simply that it doesn't make sense for Mastodon to mirror content posted from other instances. It could be CP, or it could be a copyrighted work, doesn't matter. While this is clearly an issue to be resolved, it's been overhyped and a resolution is underway. I think any court would be friendly to a defendant who mirrored illegal content unintentionally through the behavior of their software in the meantime.

[1] https://en.wikipedia.org/wiki/Legal_status_of_drawn_pornography_depicting_minors

UE @ue_ Peer Contributor

This is rather tangentially related, and may be an unpopular opinion, but I think it is silly that drawings of fictional characters who appear to be under the age of 18, nude or engaged in sexual activity, is posing such a legal issue. It's very strange and in my opinion ridiculous that such material is illegal.

Though I can understand the wider application to materials not allowed to be possessed, there isn't much that a single instance owner can do other than to shut out the whole federation aspect of Mastodon, which would be a great loss and simply centralise things again.

If the metadata could somehow be moved from the server to the client, or the client chooses to whom to subscribe and post to, wouldn't that be better? Why is it done on the server side?

Edit: it would be nice if people could please explain downvotes, so I know how to better improve the posts I write. Downvoting itself is not helpful to this understanding.

CO @ColinWright Peer Contributor

From the article:

``` So I opened up a server and started monitoring it. I invited any kind of user in with a strict “no rules” policy.

``` That's admirable, but these days it's totally naïve. Seriously, how can anyone with any experience at all of on-line communities believe that this could ever end well?

``` ... I would need to monitor what my users are saying at all times, ... goes against my hands off approach. So I’m closing the server ...

``` In today's world you either facilitate people doing some things that are utterly vile, or you monitor, constrain, curate, and censor. You can't blame Mastodon for this (although the author never explicitly said they did blame Mastodon, I feel that it's the impression they want to give).

RU @RunningDroid Peer Contributor

I think this is where NNTPchan (and to some extent imageboard culture itself) helps with working around legal issues.

On imageboards content is confined to relevant boards e.g. posts about anime would go on /a/(anime) not /k/(weapons) unless they are weapon-themed anime. In addition posting illegal/unsavory content (e.g. cp or gore) on boards it doesn't belong is usually considered a ban-worthy offence.

The way NNTPchan specifically helps is by letting the server admin decide which boards are permitted, so NNTPchan servers have the option to not handle any content related to /loli/ or any other boards that could get them into legal trouble.

HISTORICAL ERA Cloud & Mobile Era (2014–2019)
COMMUNITY SCORE ★ 67 points
TOTAL DISCUSSIONS 79 comments
TOPICS & TECH STACK: