Conference Beverages

Conference Beverages!

I recently asked a question on twitter:

And the results (raw data available in the addendum) were fantastic!

As chair for PyCon 2018 something I had been thinking about was how we can work to improve the atmosphere at events which do include alcohol service for those who do not drink.

As a non-expert, I'm going to be summarizing the responses and sharing what I learned here. Further feedback from the community is most welcome!

On a personal note much of this was driven by feedback and thoughts I'd heard from the community. The topic has been top of mind since August when my best friend undertook a journey to sobriety after years of alcoholism. I'm incredibly proud of you Jon, keep up the amazing work.

Collection of data

Yo, Twitter. That was harder than it needed to be.

I wrote some Python to take an HTML dump of a tweet permalink and dump out the basics like reply count, likes, retweets, and the tweet body.

Unfortunately, I had to go through manually to click all the daggon "show more" buttons before copying out the HTML. I probably could have written some JavaScript...

The resulting comma separated values file was manually cleaned of off-topic responses and other "meta" conversation.

Linked References

Here are rock solid references I received in response:

Summarization of responses

I wanted to do this with some Natural Language Processing, but couldn't figure it out... instead I used my own smarts to manually classify each tweet. The classified.csv file linked in the Addendum contains my work.

This is probably the largest place where my biases and ingorance muddied the waters, though I tried to critically evaluate each tweet and went back through in multiple passes to adjust.

Raw Classified Responses

Let's just count the frequency of the classified drinks in responses, ignoring likes/retweets.

Top 15, full results available in addendum, non-classified counts removed.

drink_classifier count
tea_that_does_not_suck 18
non_alcoholic_cocktails 18
sparkling_water 16
ginger_beer 12
craft_sodas 10
fresh_juice 9
low_calorie_sugar_free 8
caffeine_free 7
starades 6
just_plain_water 6
carbonated_mineral_water 5
coffee 5
root_beer 5
infused_water 5
hot_chocolate 4

Weighted Responses

Doing some super naive weighting of +1 point per like, +2 points per retweet, I am very pleased that the top responses don't change by much. It does however boost some of the otherwise infrequent classifiers... food for thought.

Top 15, full results available in Addendum, non-classified counts removed.

drink_classifier count
tea_that_does_not_suck 86
non_alcoholic_cocktails 75
ginger_beer 50
sparkling_water 50
fresh_juice 33
craft_sodas 26
low_calorie_sugar_free 25
kombucha 23
caffeine_free 23
infused_water 22
carbonated_mineral_water 21
lemon_lime_bitters 16
starades 15
root_beer 13
coffee 12


First things first

I love the passion, diversity, and openness obvious in the responses. The data collected here is a superb starting point for improving the atmosphere and inclusivity of conferences.

My motivation for starting this conversation was specifically focused on what PyCon can do to make sure that folks who do not consume alcohol feel welcomed and included at our events where alcohol is provided. Not only does this benefit your community by being more welcoming, there are certainly individual advantages to encouraging a reduction in alcohol consumption.

The thoughts that follow are therefore broken up into my take on how conferences can use the feedback here to better serve their attendees in different environments.

Don't take any of this as a promise for events I'm involved with the planning of. This is exploratory and we're going to work hard to make progress in this direction. Also, please don't flame me when your drink of choice isn't available at PyCon 2018 in Cleveland Ohio, this wasn't and could not have been a shopping list.

If you do have any dietary restrictions, we'll do everything we can to accommodate those. Be sure to note them while registering for the event and contact the resources on PyCon's registration page if things change between your registration and the conference's start.

Breakfast, Lunch, Breaks, and Hallway

Tea that does not suck

This came to the top of both result lists. It makes sense to me as there is a large bias towards the United States’ fervent coffee preference at US-based events, while tea is preferred as the caffeinated (or not) hot beverage of choice for nearly every other country.

Apparently few get this right. Tepid water, shitty teabags, and poor selection were common themes in tea related responses. Lots of folks adored the idea of large pots of pre brewed tea served hot. If that's not an option, full tea setups with boiling water, a selection of teas (including herbal), and milk/sugar are a next best bet.

Low Sugar and Low Caffeine

Not everyone wants to be wired every second of your conference, but something fizzy or just lightly flavored water can be a bit more refreshing than a glass of plain old water. It is also welcoming to health conscious attendees who simply don't prefer to take in calories via beverages.

Freakin' Water

Have plain old water available too. If you're a human, you're about 70% water. While the cost of bottled or filtered water may be a bit higher, attendees come from all over for your conference and may not feel comfortable drinking tap water in a new environment. As an aside, I was blown away to read some responses that noted that water required drink tickets in some venues...

Fresh juices

This is likely out of reach for most conference budgets, but was a very common theme. A few respondents were asked about what this means and the goal isn't "in front of your very eyes" but rather juices that haven't been laden with preservatives, additional sugar, or given enough time to oxidize.

At the bar

Conferences for all industries and booze seem to be attached at the hip. Model View Culture "Alcohol and Inclusivity: Planning Tech Events with Non-Alcoholic Options" dives headfirst into this pattern with a focus on tech and if you've stuck with me up to this point, you'll probably get even more out of that post. Go read it.

In general, all of the guidance from the previous section applies here. Pick what makes sense. From the perspective of giving a little bit more of a "treat" or value-add for events where you'll be providing alcohol, here are some responses that stood out.

Ginger Beer, Root Beer, and Craft Soda

Your bar has probably added some craft and local beers as options alongside the mega brewery options. Do the same for soda! Most of the time when someone sees name brand cola, lemon/lime, diet cola, and ginger ale on the bar the assumption is it's just there for mixers.

Going out of your way to provide some local or otherwise premium soda options at these events will indicate that thought went into these choices beyond what someone may prefer to mix their liquor with.

Carbonated water with a twist

Almost literally. Many folks noted that a fizzy low sugar and/or low caffeine option would be very welcome. Suggestions ranged from just soda water with a lime to more elaborate mixes of syrups, citrus, and carbonated water. I learned about a thing called the "Lemon-Lime Bitter" from an Australian contingent of respondents. But note: watch out on what kind of bitters are used for these options as most contain alcohol.

Non-Alcoholic Cocktails

By far the largest set of responses that seemed to apply here were for cocktails that do not include alcohol. Low and no alcohol cocktails are a growing trend as evidenced by linked newspaper articles. Adding some unique options for cocktails that don't include alcohol absolves your attendees of the need to negotiate with a bartender on a drink that suits their needs.

This is a strong option as it shows direct and concerted thought for attendees that don't drink alcohol.


I hope that other conferences might be able to take something away from this post. It was a bit of an undertaking to pull the data together, do something with it, and write up. That said it was also incredibly enlightening for myself in helping to prioritize some direction on first steps we can take to improve this aspect of the conference as PyCon 2018 in Cleveland comes together over the coming months.


This Gist contains all of the raw data on the responses as well as the code used to process them (excluding my manual classification).