Hacker News Reader

Comment by quanto | Hacker News Reader

There are many mushroom varieties that look similar but with completely different dietary properties, some being downright toxic. Many years ago, an agriculture research team approached me to develop a machine learning classifier to determine which mushrooms were safe to harvest and eat -- this was meant to help rural farmers and foragers. Unfortunately, the project didn't go anywhere AFAIK.

I wonder what the state of the art on this front is. Surely, someone must have come up with some solution (machine learning or not)?

View on HN

There have been unusually many mushrooms around our house due to an extra wet season, which is making me want to try picking some, but I'm terrified of the risks. Maybe it's just one of those things I should let slide.

Foraging is a great hobby as long as you approach it with respect. With mushrooms, you start by learning a few foolproof species that have no close lookalikes, and then you learn species by species from there. After a few years of doing it regularly, you'll be pretty comfortable IDing (with a field guide, and occasional help from a good Facebook group -- most regions have a good one, and you need one that's fairly close to you because they'll know the ones that grow near you). I started getting interested around 2013, but didn't eat a foraged mushroom until 2016. I've eaten around 30 different species that I've foraged at this point, all from my area (most within a mile and a half of my house).

Nope. Nope. Nope. No way is a mushroom from outside passing my lips. Good god they all look the same and even if they don't - are you 99.999999999999 percent sure?

Nitpick: a healthy young person has ballpark 1 in 10 thousand chance of dying per year. So it's enough to be 99.9999% sure because extra nines don't really make your life safer.

I'm personally willing to throw a couple of 9's on that figure to avoid dying the mushroom death.

This is one reason I generally advise beginners to learn the few deadly mushroom species alongside any edible ones they're learning, and why some edible mushrooms are considered to be for more advanced foragers because of the possibility of confusion for beginners. But there are many species of edible mushrooms with zero toxic (much less deadly) lookalikes, which is where people generally start.

Agree. Better safe than sorry. If you don't have a minimum experience in mycology don't do culinary experiments.

Yes... and -- the minimum experience needed to identify, say, chicken of the woods (Laetiporus spp.), chanterelles (Cantharellus spp.), or lion's mane/bear's tooth/coral tooth (Hericium spp.) is very, very low.

If you learn just a few things about mushrooms, they stop looking the same. Also, there are only a handful of deadly mushrooms in any given region, and you can learn those specific species as well. It's not that hard to get to 99.9999999% sure that you're eating an edible species.

An old Check proverb says something like: all mushrooms are edible; some of them only once :-)

That one and "there are old mushroom hunters and bold mushroom hunters, but no old, bold mushroom hunters" get told about once per day per thousand people in any online mushroom group.

Like many other things, finding a mentor is really the way to learn. I have gone picking with people who have done it for years and not only are they much better at finding 'the good ones' than I would be (even if I was confident I had the right species), they knew what to avoid.

my makerspace in victoria, bc runs classes on dna barcoding for fungal identification. About $12 per id when all is done.

Reply on news.ycombinator.com