Introducing Highlights

They&#x27;re doing it based on engagement? Great.. it&#x27;s just going to surface the jokes. Seriously, there&#x27;s a lot of important information that passes through our slack channels, but, for the most part, we don&#x27;t &quot;engage&quot; with it in the way they are measuring. We don&#x27;t react, reply, click on, or share the important messages. We just respond to them in the normal flow of conversation.<p>The jokes, wit, and quips on the other hand... those get reacted to. So there&#x27;s a good chance those are going to be the highlights. Which, while fun, is a complete waste of screen real estate and isn&#x27;t going to help me be more productive.

At the risk of sounding geriatric, I think Slack is too complicated. The straw that broke the camel&#x27;s back for me was threads. I&#x27;ve not yet seen a thread solve a problem that a channel would have otherwise had, but I&#x27;ve seen plenty of confusion from the added orthogonal dimension and the general orphaned nature of messages in threads. Now, as is customary, they&#x27;re putting the training wheels on their too-complex product.

&gt; Predicting which messages are most likely to be important is a challenging technical problem<p>I think this is much better solved as a social problem. Use mentions sparingly so they have value and have company conventions about what&#x27;s expected to be read&#x2F;skimmed&#x2F;ignored.<p>Seems like a case of fitting a solution to a problem.

Slack doesnt need anymore features. Just let me chat and stop making me learn new ways to do things.

they should make the search usable first

They use sampling to &quot;preserve a fixed ratio of positive to negative examples.&quot; I wonder what the ratio is... isn&#x27;t the technique w&#x2F; class imbalances to up-sample&#x2F;subsample to get a 50&#x2F;50 ratio so your classifier generalizes?<p>On the topic of feature engineering: I think the use of those inline-emojis on a message are a good indicator of importance. That could be another feature for the logistic regression.<p>Overall seems like a fun modeling challenge. It&#x27;s nice to see people not just throw deep learning at the problem...

if this is anything at all like &#x27;twitter highlights&#x27; it will be cumbersome and in the way.<p>Especially if I get prompted to read them- I hope they are configurable unlike so many other components in slack too.

I wish Slack would just fix core functionality like badges (if I open the room they should go away) and notifications (if I read the message on one platform, don&#x27;t send me notifications for that message 5 minutes later on another platform). Those are the features I want.

Any good Slack alternative for communities? We have a big community and are looking for a chat client but there is no way that we would be willing to pay $5&#x2F;user when we have up to 1000 users.

I don&#x27;t care about this really. All I want is just a fluid, fast desktop app that won&#x27;t make my MBP overheat. Is that too much to ask? =(

Every time they introduce a new feature, people at work start using it so I have to try to understand it. I still find the threading confusing.

They should decrease memory usage first for desktop app. I hope you can disable this highlights.