Ask HN: What are some bad engineering decisions that screwed startups?

[marwann]

I'm looking for startups that failed because of a big engineering mistake, be it technical or in the management side, and I can't think of many famous examples.

I'm thinking about some reasons: - Being dependant on external services/code, eg. when an author removed some modules on NPM that caused thousands of websites to break - Bad hiring decisions in tech teams - Trying to patch bad code instead of starting over - Focusing on research (or code cleanliness) instead of finding product market fit

Maybe Google wave that had a buggy interface (though I suspect this wasn't the main problem), HP Touchpad that went out in a rush and was clearly badly developed.

Any clue?

[epc]

You could argue Friendster lost a lot of its early, first mover, momentum to MySpace and Facebook because of its initial implementation in Java JSP. By the time of its rewrite in PHP, MySpace had become the "in place", followed fairly quickly by Facebook (while MySpace foundered under Newscorp ownership).

You don't hear about the failures because they rarely get enough traction to be popular (and if they got traction then typically investors will pour money in to keep the startup afloat).

Also many survivors of such startups are under NDAs which mostly serve to protect the management teams.

[marwann]

Didn't know about Friendster's backstory, thanks. Did JSP make the site so slow it made people flee to Myspace?

[epc]

I wouldn't say people fled to MySpace…friendster had this initial surge of users, the site become unreliable (and once you set up an account and got over the initial flurry of fun, there was no "there" there. You could connect to people, you could send messages, but IIRC there was no "timeline" or "newsfeed"). MySpace launched within a few months after Friendster (allegedly inspired by Friendster), had more features, and most importantly didn't crash as nearly as often as Friendster. MySpace also was more "creative" in a GeoCities sort of way, you could tailor your page in a variety of ways, and it appealed to teens (who would go on to create Facebook accounts within a couple of years and shift their loyalties).

[blodovnik]

Twitter didn't fail but it was on its knees for a long time due to poor architecture and technical capability in its early years.

[marwann]

Thanks. Do you know more about the actual flaws of the architecture in the early days?

As I recall, they used to be hosted on Akamai and in colocation centres, but I don't know much about the rest.

[epc]

Initial Twitter service was built using Ruby on Rails, shifting to Scala on the JVM around 2011-2012. See: https://www.infoq.com/articles/twitter-java-use/ https://carlosbecker.com/posts/twitter-drops-ruby-bullshit/ https://blog.twitter.com/engineering/en_us/a/2011/twitter-se... for a sampling of explanations.

[marwann]

Thanks a bunch!