Elsie Nakamura

Four conversations a month from a channel you don't enjoy isn't a failure, it's a slow channel. My mistake was declaring a channel dead at month one. My second mistake was staying on one for a year out of stubbornness. Give it eight weeks with exactly one variable changed - I'd change the message, not the volume - then decide.

Whichever you choose, make sure a slow endpoint cannot stack requests. With a 5s interval and a widget that occasionally takes 9s you want the next tick skipped, not queued behind it - both libraries handle the in-flight case, but if you have wrapped anything in your own effect you will have reinvented a request pileup. Ask me how I know.

I would push back a little on going to the shop first. Email or use the contact form on the manufacturer's site, because the brands frequently send a voucher or a refund for a first-time buyer who did not get on with it, and they would rather keep you than win the argument. Two brands have done that for me and one sent a coupon worth more than the bag.

The 3 hours will not be 3 hours. Writing an issue is maybe 70 minutes of actual writing and then 90 minutes of second-guessing the subject line, fixing the layout in the editor and staring at the open rate afterwards. Budget for the staring.

Leave the one way valve alone and keep the bag sealed rather than trying to speed degassing by opening it. Oxygen staling is a much worse and faster problem than CO2 is, and it is not reversible.

Five weeks in, search is the wrong first channel, and I think this question is partly a way of not having to talk to people. Write if you enjoy writing, it compounds, fine. But your first ten customers this year are not arriving from search, and planning as though they might is how people end up with 40 posts and no product feedback.

Ring of tabs rather than one under the crown, got it. How deep, mine is basically at the roots.

You can get almost all of it out without buying anything. The plunger screen is not what does the filtering, settling is.

What I do:

• pour all the water, wait 4 minutes
• break the crust with a spoon and scoop off the foam and floaters, they hold a lot of the fines
• wait another 5 to 8 minutes without touching it, so the fines settle
• push the plunger down only until it just touches the surface, do not press it to the bottom
• pour slowly and stop when you see the cup start to cloud, leaving the last 50 to 80ml behind

The extra waiting sounds ridiculous but the coffee is still hot enough and the difference in the cup is dramatic.

Worth writing down what the sandbox is for before building it. Mine started as security and turned out to be mostly reproducibility - the value was that every run started from a known state, so a failing test was the agent's fault rather than my machine's. That framing also tells you what to cache and what must be fresh.

Also budget about 30 dollars of the 400 for the small stuff, a decent tamp that fits your basket, a scale that reads to 0.1g, and a bottomless portafilter if one exists cheaply for your machine. Those three change more shots than the last 100 dollars of machine would.