E056: Learning when NOT to build that cool new feature
Sometimes the better answer is to not do anything at all. If you can't prove the value of a feature or that the effort to achieve that value is worth it, saying "no" to the idea is often the best course of action.
We also look at scenarios where Product Managers and Engineers can benefit from being users of their own software, as well as the benefits of having competitors you can "borrow" inspiration from.
Join the discussion on Reddit: https://www.reddit.com/r/AcceptanceCriteria/
And on the Discord: https://discord.gg/2Tyj8H9MFF
The post E056: Learning when NOT to build that cool new feature first appeared on Acceptance Criteria.
-------- Â
E055: Mark Zuckerberg and Meta bend the knee to bigotry
Kevin gets a little spicy in this bleep-filled episode discussing the latest changes to Meta's policies around fact-checking and hate speech. Trans rights are human rights, and bowing to pressure from Trump and Republicans on letting people use slurs and lie to each other poisons the well for the rest of us.
Content warning for difficult discussions of trans rights, vaccines and autism, and a variety of things Meta has now decided aren't worth blocking.
Join the discussion on Reddit: https://www.reddit.com/r/AcceptanceCriteria/
And on the Discord: https://discord.gg/2Tyj8H9MFF
The post E055: Mark Zuckerberg and Meta bend the knee to bigotry first appeared on Acceptance Criteria.
-------- Â
E054: Is Deep Seek’s AI the ultimate disruptor?
Chinese LLM company Deep Seek upended the AI market this week with a new AI model that is comparable to ChatGPT and other AIs but appears to have cost barely $5 million to create.
At one-twentieth the cost of Open AI, Meta, and others' models, it begs the question: what the hell is the US market for AI doing asking for $500 BILLION right now??
Also, we look at Google's quantum computing breakthroughs and debate whether physicists actually know what they're even talking about. A conversation we're both equally capable of handling :)
Join the discussion on Reddit: https://www.reddit.com/r/AcceptanceCriteria/
And on the Discord: https://discord.gg/2Tyj8H9MFF
The post E054: Is Deep Seek’s AI the ultimate disruptor? first appeared on Acceptance Criteria.
-------- Â
E053: Is there value in fixing every bug?
We tackle some more Reddit questions, including someone who's looking for ways to track their teams' mood, how to find a mentor, how often to constructively criticize your agile process, and whether fixing every bug is bad, actually.
Join the discussion on Reddit: https://www.reddit.com/r/AcceptanceCriteria/
And on the Discord: https://discord.gg/2Tyj8H9MFF
The post E053: Is there value in fixing every bug? first appeared on Acceptance Criteria.
-------- Â
E052: Why don’t front-ends look like the superior UX designs?
From understanding why UX designs don't always translate to the actual front-end code, to finding the creativity and joy in maintaining other people's code, it's another Reddit grab bag.
Join the discussion on Reddit: https://www.reddit.com/r/AcceptanceCriteria/
And on the Discord: https://discord.gg/2Tyj8H9MFF
The post E052: Why don’t front-ends look like the superior UX designs? first appeared on Acceptance Criteria.
Listen to Acceptance Criteria, The AI Daily Brief (Formerly The AI Breakdown): Artificial Intelligence News and Analysis and many other podcasts from around the world with the radio.net app