Effective Data Practices for Coffee Roasters and Other Recent News

While the latest release of Typica was only a point release by version numbers, it included a lot of new functionality that I thought would be nice to promote more widely, so I put together a press release and sent it to a few outlets that I thought might be interested since they've done new product or product update announcements in the past and that's exactly what this is. The responses to that were generally along the lines of, "congratulations, looks great, we're not going to mention that anywhere."

The exception was Perfect Daily Grind, which sent a counter-pitch, asking me to write an article on the sorts of data that it's useful for roasters to collect. I decided to take the opportunity to do sort of a high level survey on data collection methods, evaluating effectiveness of data collection practices, and going beyond just talk of profiles (which are, of course, still important to capture) and into production, lab, and sales data. You can read it here. Before it went to online I did tell them about the extra s and that the title they made up for me wasn't accurate, but it seems they didn't bother to make those corrections. Otherwise, I think they did a nice job with the editing.

Everything in that article can be expanded on substantially, but if I did that it wouldn't be an article anymore. It would be a book. I already have too many books in progress that I need to finish writing, but I guess that might end up being another one at some point.

Tomorrow I'll be getting back into town. I've been away from home teaching classes at Roasters Guild Retreat and finishing up some of the SCAA heritage material so I can have as much as possible transfer into the new curriculum (I've also been busy helping out with developing that material). Retreat was a lot of fun. I talked with someone who listened to me talking about some work I'd done in 2015, repeated a profile tweak that I'd found to maximize sweetness, and found that his customers unprompted started commenting specifically on the improved sweetness in his coffees. For those curious, the tweak was to take the portion of the roast from the start of yellowing through the green to brown transition in 100 seconds (equivalent to an average rate of change of 9 Fahrenheit degrees per 30 seconds on my roaster, but your measurements may differ so do your own math). Faster or slower while keeping the profile outside of that range the same as shown in profile translation analysis seems to reliably reduce perceived intensity of sweetness across a broad range of coffees. I'm not saying your profiles must do that. You may want to take that range faster or slower if you're willing to trade some sweetness for an improvement in some other quality.

Excessively slow drum speed experiment.

Another highlight of retreat was this amazing roast in the heat transfer class. I've never before managed to get a roast to take so inconsistently that some coffee in the drum was still green while other seeds were fully in first crack at the same time. Eventually bean temperature readings started to decline, but by this point between the extreme darkness of some of the seeds and the excessive chaff build up in the drum, it seemed likely that I could have had a fire if the roast weren't stopped.

Tomorrow I should also receive some new hardware from Phidgets. They've recently released a new product line and a new version of the library used to communicate with that hardware. Since the new library is no where near source compatible with the previous version, Typica will need new code to add support for this. I'm hoping to get a 1.8.1 release out to add this within about a week, but until then, new installations using the Phidgets 1048 should continue to use the previous library version and not the new one. I want to give a shout out to their developer support for providing the hardware I need to test required changes in Typica against. Without that, it would take considerably longer for me to get around to writing that.