In episode 8 of Unintended Consequences, Heidi Waterhouse and Kim Harrison speak with J. Paul Reed of Netflix. They discuss resilience engineering at Netflix, the emergent nature of systems, surprises at scale, and the future of conferences.
About the Guests
J. Paul Reed is Senior Applied Resilience Engineer at Netflix. He is a recognized speaker on DevOps, release engineering, and operations complexity. J. Paul has written articles for O’Reilly, DZone, and Atlassian and is the author of DevOps in Practice.
Heidi Waterhouse: Today. We have with us J. Paul Reed, who is currently making Netflix work.
J. Paul Reed: I got a lot of help in making Netflix work, but the way I actually kind of like to think about it, is that I help the people that make Netflix work, better understand all the ways in which they make Netflix work, that may not be super obvious to them.
Heidi: I guess that makes sense, because there's the things that we do and then like the meta understanding of the things that we do, which we can't really do when we're in an emergency or reacting.
Paul: Yeah. So, for folks that aren't familiar, I'm a Senior Applied Resilience Engineer at Netflix and the implied is very important, but what's interesting about that is--
A lot of times it goes back to this age old question around, we always post-mortem when the site was down, but will we never post-mortem when the site is up.
And so helping people understand how they go about doing their work, how that makes actually the system work, how it makes it "safer" or more resilient, so then we can do more of that.
As opposed to just being reactive in a incident context or an outage context.
That's what I spend a lot of my time thinking about and helping folks on that journey at Netflix, and it's a lot of fun.
Heidi: Right, and I never remember the actual name of your degree.
So I just tell people that you have a degree in disasterology.
Paul: I love that. I'm going to have to start using that. So the degree is in Human Factors and System Safety, it's from a University in Sweden, Lund University.
Although I had a friend once joked that it was the, "Alspaugh Swedish School of Massage," actually, was what he said, but I kind of like the Alspaugh Swedish School of Disasterology, actually there you go.
That's what it should be.
And it's funny, by the way, so for folks that aren't familiar with that program, I was just telling someone about this and a little bit of the history I think is useful to talk about.
The reason we call it they Alspaugh Swedish School of whatever is, because he was the first software person to go through this program.
And what's unique about the program is that was originally designed for air traffic controllers and pilots and doctors and nurses.
And there's a lot of maritime shipping and maritime gas and oil extraction.
So, people driving boats. In fact, Lund University has a boat simulator, I'm sorry, no that was Chalmers, which is another Swedish University.
Anyway, we got to go play with the boat simulator, which was a lot of fun.
It's like an aviation simulator, like you're on the bridge of the boat.
You get to steer it, we got the bump the-- I think we actually smashed the boat into a bridge, right in front of the Sydney Opera House was where they set up the simulation-
Heidi: Did you drift it through a canal.
Paul: You know, funny you say that, now that's the whole meme and whatever, but so this is a bunch of like--
We made sure when we were in the simulator that there weren't any maritime people there.
Because we didn't want the advice, it was all like, computer people and doctors and stuff.
And so we were like, "Let's take this huge container ship on like max power."
And then it was a game of chicken with the coast, because we were trying to see how close we could get to the coast.
And so the best part is, we actually did crash into a cruise ship, but it's not the crash the way, because you're on the water, and so the boats actually just came together and kind of bounced off.
Anyway though, he was person who went through this program, and I was the second person and then Nora Jones and some other folks were in the cohort after that.
So I think we're up to all told maybe 10 software people in the cohort after me and the one after that.
And Nora Jones tells this great story.
I love this story, where they're doing intros on that first day.
And this was before I was in the program, before I worked at Netflix, but Nora was at Netflix at the time.
And so it's like, "What do you do?"
"Well, I'm a trauma surgeon."
"What do you do?"
Well, I'm an air traffic controller at London approach."
"What do you do?" "I'm a pilot for SAS," or whatever.
And they're like, "What do you do Nora?"
And she's like, "I work at Netflix."
And she says, they all laughed and said, "No, really what do you do? And why are you here in the safety program?"
And she would say, "No, I really work for Netflix," and had to walk them through and explain with her colleagues or her classmates, why she was there.
But I love that this has become a thing.
That human factor, system safety, and software has become more of a thing.
Because it's funny, we talk DevOps traditionally had that culture and empathy and that whole component.
And it's really taking the tact that the social parts of the system, the socio and socio-technical system actually matters.
And I think for so many years we just kind of thought, "Nah, if we write enough Runbooks and automate enough things, we can kind of ignore the dynamics of the people and how they contribute."
And that's a major shift, that it's the people that makes the whole thing work, and you need to pay attention to that.
Heidi: Human factors are so important.
And when we're talking about DevOps and surprises that scale, the whole point of scaling problems is that you can anticipate a lot of them, but some of them are just unimaginable until you get there, like who would have thunk it.
Paul: Well, what we often talk about in that context, we talk about the emergent nature of systems.
And so when you talk about scaling and the surprises and the challenges, it really is a lot of the emergent aspect of a system.
I was just in a meeting recently where we were talking about a reorg.
Netflix is famous for reorgs. And somebody actually said it this way, and I really liked how they said it.
They said, "At Netflix, we don't want to ever fall into the trap of shipping the org chart, shipping a mirror of our org chart and software."
So they will reorg pretty often, not the entire company, but parts of it to better reflect what the customer needs and the product needs and all that kind of stuff are. But we were talking about, when you start doing that, you can find situations where you may have incidents because of a reorg. Because the expertise got moved around or something happened there.
So that's an example of an emergent thing that when you talk from a scaling perspective, if you believe that's a good way to be comfortable with reorgs and be okay with that, as your company grows, there's going to be more of them spread out across the company.
They'll just be more common because of the size of the company.
And that's one of the vast numbers of emergent things that you end up having to pay attention to.
And I think the one thing I'll just say real quick, I think that emergence happens whether or not you want it to happen.
It's really around, are you paying attention to capture what you can learn from that?
And organizations are really good at paying attention to pain.
So that's why incidents are the tip of the spear for that sort of thing.
Because that they're painful to the org and a lot orgs try to figure out how much an incident costs, and that's an internal debate.
Can you measure the cost of an incident?
And so that's where they experienced that emergence, but you can experience it in other ways too, and if you have folks that are sort of dedicated to paying attention to that, I think that can be super useful.
Heidi: Yeah, I think we think of emergence and the emergent properties of systems as a sort of thing that happens to us and as an emergency, it's not an accident, these words are very closely related.
Also, the properties of a system are responding to the external pressures to make the system the way it is.
All systems are responding to positive and negative pressures that we may not be paying attention to.
And I think the pandemic is a really interesting example, so Netflix has all of this great streaming stuff to offer people who are now housebound forever.
But how do you film new things? What do you do?
Do you go out and contract for like every Gilligan's Island episode ever in the hopes that people will somehow bond with Maryanne?
Kim Harrison: Wait, can we back up for a second and recognize that when we all went home, Netflix was still streaming in a stable way.
We've had issues with a lot of other things, but Netflix stood up.
Paul: I can talk a little bit about that.
It's funny, I was chatting with a colleague, Lorin Hochstein, and he was talking about, there are certain companies that if you were inside the walls, when the pandemic happened, you're going to have a lot of interesting stories to tell.
So Netflix is one of those. Zoom is another one of those.
And I'm sure, the common ones, AWS and Google, I'm sure they have their stories as well.
You know, Facebook, people were probably on social media a little more often, but before we talk about that, one of the things I actually wanted to mention and just go back to that idea that you brought up of emergent and emergency, and how those are related.
One of the things, when we look at those properties, the systems, and one of the things from a resilience perspective that we're interested in trying to do is, we want to dampen the emergent properties that are negative, that we say, "Oh, we didn't expect that, and that's not great."
So we want to dampen that. It's not like a switch.
You can't turn it on off, but you can do things to dampen it.
And we see that a lot in incident follow-up.
We had something that broke, and we might want to look at the emergent things that contributed to that breaking.
Heidi: You all, can't see it. But every time Paul says caused-
Paul: It's air quotes.
Heidi: There's air quotes around it, because he is literally incapable of an unironic cause or root cause attribution.
Paul: Well, and the funny thing is when anybody says root cause including me, my face like scrunches up, and I cringe a little bit on the inside.
But the thing we don't talk about a lot of times is, we can dampen the bad things.
Great. But how do we amplify the good things?
We don't often talk about the amplification aspect of emergent properties, because we're so interested in, make the pain stop, let's dampen that, that oftentimes we'll miss stuff that we should be amplifying.
So Kim, your question I got to ask before I tell this story, I'm going to charge you with a question.
Kim, what did you watch during the pandemic on Netflix the first couple of months?
What was your keeping your attention?
Kim: It's kind of heavy, but I rewatched The Good Place, because I knew that the final season was going to come out, and so I felt like--
I live alone in a small studio apartment in downtown Oakland, and this is awful being completely cut off. I don't have family in the Bay area, and I needed something positive.
So I rewatched it as a lead-up to the final season that dropped, I think that was in April.
Paul: Yeah, it's so good. I love that show. So good.
Kim: Yeah, So good.
Paul: So good. Heidi, what about you?
Heidi: So, my daughter and I watched through Call the Midwife, yet again.
I think it's her fourth time through, deeply addicted to Call the Midwife, and so she made me watch it with her.
Paul: I skipped around a lot. I had finished up The Good Place, I think.
And again, it's funny, I'm actually rewatching it now, because I had a friend who didn't see it. So, we're watching it together.
Heidi: There are so many little jokes you don't get the first time-
Paul: It's very, yeah-
Kim: It's amazing.
Paul: Yeah, It's some dense writing.
Heidi: Restaurant names.
Paul: Yep, for a while I went to a very dark place, and I think all of us experienced a little bit of that in the pandemic.
So, for whatever reason, I leaned into the dark place, and I watched a lot of Black Mirror.
I love Black Mirror, but then I watched half of it, and then I was like, "I think I need, I got to go watch Schitt's Creek," which I had not watched.
I made my way through that, and I made my way through Kim's Convenience, which are both great shows.
Kim's Convenience is amazing. I love that show.
So the thing that I will tell you about COVID-19 at Netflix is, or when the lockdowns started happening in March and into April, so I'm on the CORE team, which stands for critical, operations, and reliability engineering.
So we are kind of the top level incident management group.
And so we often say, and my colleagues that do this are amazing, we say, "We hold the pager for Netflix."
By the way, I told them in the pandemic, my mom got this new hobby where she would like, anytime that she would see Netflix is down for like a friend on Facebook, they'd say, "I think Netflix is down." She would actually text me and be like, "Is Netflix down?" And so I was telling my colleagues, we've got multimillion dollar monitoring stuff. Let's just turn that off and use my mom. She'll text us when it's down.
It's fine. But so, we ended up treating COVID as a CORE managed incident.
There was a lot of work around the company to make sure, as you might imagine, people were home and they were streaming more.
And so there were a lot of really great work done, folks were really attentive to their own systems, like trying to work through the additional load, and what are they going to do about it?
We had a daily, in fact, on the calendar, it was every morning, I think at 10 in the morning and the meeting for the incident responders, and I was one of those for the COVID managed incident, the meeting was called, and now your moment of COVID-19, if you remember from the Daily Show, your moment of Zen?
So, we actively managed it.
And there were a lot of things that we sort of helped to coordinate, make sure that information got around the organization.
Are open connect team that delivers the video bits.
They did some work in, and you can Google around this, but there were certain countries that were heavily impacted.
I think Italy was one of them early on where the government was kind of like, "Hey, people are watching a lot of Netflix, and it's causing problems for the internet in the country. Can you fix that?" And so we-
Heidi: And you're like, "Our infrastructure team will just fly in to your COVID ridden nation and lay some more fiber. Sure."
Paul: It's kind of like, "Listen, just take the ethernet cable and plug it into your neighbor's ethernet jack. It's fine."
Like, "Plug it into Greece's ethernet or something."
But so, they get some really great work where they were able to detect kind of, network hotspots, and then change the bit rates at which the videos would play to help make sure that those more critical infrastructure in terms of stuff on the internet in their country was not affected.
So, there was some tremendous engineering work done there.
The other thing that I don't think we talk a lot about is--
You mentioned that the studios halting productions, and there certainly was that and Netflix was one of the companies that created really early on with some hardship funds for productions that are Netflix originals, where it's like, "We literally can't film."
" So, what do we do, because now we're not paying you, because we're not filming. So what do we do about that?"
So they did a number of-- They called them hardship funds for productions that were affected by that.
But the other thing that you have to remember that is a story, and I've told this story a few times about COVID, but the more I think about it, there was such a tremendous weight I think of everybody doing that work at Netflix from the standpoint of--
We knew that we were a place that people were finding escape and moments of joy in a really bleak situation.
And we were really glad that we could help in that way, but we are also all humans, and we're also all working from home and figuring that out.
And Netflix is famously not remote.
And we're actually reexamining that as a company now, about, is that the right decision moving forward, coming out of the pandemic.
But you know, we had to struggle with all that ourselves while also trying to be there for everyone else.
And so this is not a boo-hoo, woe is us story, but it really is--
Heidi, we were talking about socio-technical systems, the COVID story is largely, can be a framing about technical systems.
How did Netflix make sure that the internet didn't crash, or how did we manage some of the resources that were required to make sure that we kept the system stable, all that stuff.
That's all technical stuff. But then a lot of the work that I did was really around socio-guidance.
I wrote a memo for the organization around burnout, and it was a curated sort of write-up of all of Dr. Christina Maslach's work at Berkeley.
Yeah, Heidi, you're shaking your head, you know-
Paul: Exactly what I'm talking about.
That went through the org, because it was like, "Listen, in the April, May timeframe, you need to be aware that we may see burnout. We may see it in different ways."
And, I can tell you on our team, we had some folks that were pretty impacted, pretty early on, like within two to three weeks of the lockdown, they were pretty impacted.
And impacted by like, I don't have kids, they had kids.
So they were with their kids and that was a whole thing.
I know that I was impacted, but I was pretty good for the first four to five months.
And so it's really interesting to think, and folks listening to this, you might think about we your colleagues, we were kind of starting to draw a little faux graph of like, "I can take the pressure and then do you fall off a cliff."
Do you degrade kind of linearly, like how do you actually, as a person degrade in the system?
We can think about it in distributed systems, in the same way, how do you show that degradation of performance?
And some folks again, can take a beating and keep on ticking.
But then at some point they hit a wall, and it's a total shutdown, and they need a month off or whatever it might be.
Heidi: Right, and mine is an intermittent failure.
Heidi: Like, "I'm doing fine. I'll tell you, I'm doing fine, except I dropped like four balls last week, but I'm doing fine."
Paul: So, I'll tell you this, this is what I realized about myself.
I realized that I, in COVID times, even though we couldn't do anything.
I needed to take a week off every five to six weeks.
And if I didn't get that, and this is where I love working with my team, they're a great group of folks.
I told them, one of them basically said, "Listen, about four or five weeks in, Paul, you get a little crunchy."
And I'm like, "What do you mean?"
And they said, "Well, you remember that Snickers commercial where it's like, 'Dude, you're playing like Betty White.'And it's like, it is Betty White."
And I was like, "Oh, I get what you're saying."
So now actually, when it's four to five weeks, and I should put a week on the calendar, just be away.
My coworkers will actually say, "Paul, do you need a Snickers? You're playing like Betty White out there."
So, that's something that I cherish with my team that we have that kind of space where we can do that.
And I've done that for them actually, where I tend to do it is actually more in incidents, because I've done the thing where, there's an incident that starts at 11:30 and then I don't hold the pager.
I'm not on the on-call rotation, but I'll come in at 2:00 AM and be like, "You know what? I think you should go to bed, the incident's stable, you should drop the debugging you're doing. It's okay."
And so that's kind of how I get them back. I tell them that they should go to bed, not have a Snickers, go to bed.
Heidi: So, I had another question, in your not Netflix capacity, as a conference runner, where do you see us going in the next year?
How are we going to scale that out?
I cannot believe that AWS is planning to do re:Invent, which just seems like the worst of all possible worlds, but maybe that's a scaling thing that I am just not anticipating correctly. What do you think?
Paul: So I think a few things come to mind, so Heidi, you're referring to re:Deploy and you came to the last one. You've been to both of them, right?
Heidi: Yep, I think.
Paul: Yeah, and so I totally want to do it again.
For smaller conferences, the real question is, can you even schedule them now?
Because AWS, if they have to cancel, that's fine.
Whereas, if I have to cancel, that's like, I'm paying for that personally.
And that's a big pill to swallow, but it's super important.
And here's the thing for everybody who's been to re:Deploy, I actually want to hear all the COVID stories.
Like, "Do you want to talk about resilience? Let's figure that out."
But I agree with you, I don't know what it looks like.
And I have had people asking me, so we usually do it in the September, October timeframe.
Actually, it's funny, it's probably too late to even get that started, because all the planning and stuff you have to do.
But we, as a country, we keep making progress and then people get eager and they go out and then the rates go back up and we just kind of seesaw back and forth between these two.
And so I think that's what makes it hard.
Here's the other thing I'll say about conferences in general though, you want to talk about scaling that back up.
I was walking through The Castro here in San Francisco the other day, and I am an introvert, but I love being at conferences.
I love giving talks on stuff that's important to me.
And so I don't consider myself as somebody who has an issue with being around lots of people.
I like that, even though I'm an introvert.
But as I was walking through, I realized I'm having a little bit of anxiety and it was around, just there's lots of people there, a lot of them aren't wearing masks, because they were out on the sidewalk where there are tables.
So, they were eating, and so that's why they didn't have masks.
And I remember having a conversation, it might've been with Emily Freeman, and we were talking about like, it's going to be a little weird, because there's a lot of people that I'm going to want to--
I'm going to see them, and I'm going to want to run up and hug them.
And that's the thing that like, I'm going to be so excited, but you should ask before hugging random folks. And I can also-
Heidi: Well even folks you know and love.
Paul: Exactly, yeah, 100%.
That's actually my point, Heidi, right?
Is I feel like when we go back to conferences, there's just going to be a lot of, you want to talk about emergent stuff.
There's going to be a lot of emergent feelings.
That's what I think about, is how do I-- Couple things, what can I try to plan for, like wanting to hug everyone, so that I make sure that I ask, "Is that okay?"
But then also we're probably going to make mistakes.
And so it might be worth, not in just about the hugging thing, but other things.
And so, what's the grace that we can give each other and ourselves.
And where, by the way, does grace not play.
In terms of, there's a line, if something happens, grace is not the right answer.
So those are all some things I'm actually thinking about, again, when we get back to not the technical aspects of running the conference, but the socio-parts of it.
And I actually think about this going back to the office.
We were just recently talking about someone who is-- She is so good at an incident.
She's a part of the incident, she's amazing.
And I kind of made the joke, "I just can't wait for her to yell at me in an incident again," because she is so good.
And everybody that I was chatting with was kind of laughing about that.
So, there's that aspect to it too, where it's like, you're going to find out things that you missed, that you didn't know you missed, and the person I was talking about, I really miss her participation in even something like an incident, that's not good, because she's just so on point and great at her job. She's such an expert.
So yeah, conference wise, there's lots of things I think to think about, but let's put it this way.
The conferences will figure themselves out.
I know you've been involved with organizing conferences.
We were talking about Bridget and DevOps Days Minneapolis, I think DevOps Days, will figure it out.
I have no concerns about that, and I don't have concerns about the socio-side of it either.
It's more, that's where I find myself thinking about that aspect more, and what can we game out versus what's going to be emergent.
What's going to surprise us, right?
Paul: I'm probably going to cry a lot, actually, when I see folks. That'll probably surprise me.
Heidi: It's going to be so weird for us to remember how to relate to other humans.
And I actually think that there's enormous potential for people having forgotten their company manners around gender and sexual harassment, and I'm a little worried, honestly, that people are going to leave their homes and just forget everything they ever knew about acting professionally.
Yes, we have been wearing pajama pants for two years, that does not mean you can touch me like that.
Paul: So, it's interesting that you say that, because I think there's that, but it's funny, I think we are starting to see that just even in the public square.
I think about people riding on airplanes, and are we going to mask, are we not going to mask, like all those questions, where I'm like-- I was just talking with a friend, it's weird thinking about being in an airport again.
I'm thinking about being in an airport security line.
Do you think that's going to be six feet?
And I even see it now-- I'll tell this story, and it's a hard story.
I got my first COVID shot, and it was an interaction that was, "Yeah," but it was difficult, because there was another person, there was a couple there actually, and it was a mom and a dad and they had a little kid that was not masked and was running up and down the COVID line.
The line for vaccinations, and the little kid was talking to a woman that had a dog.
And so the kid was interested in the dog.
And so I kind of finally asked this, "Is this really appropriate?"
You know, "That you should be doing this."
And it was funny because their reaction was like, "Well, the kid is two, so we don't have to mask her."
And I was like, "No, I'm not asking you to mask her. I get it, two year olds don't really understand masks, I get that. But what I don't understand is why," and this kind of comes back into system safety, "your making a risk decision for all of us, without asking us, is it okay to let my child run up and down the line without a mask on."
And I get that-- And it's funny, I had a conversation with a friend of mine who was actually there with me to get the shot, and I understand it's hard.
Heidi your a mom, right?
Paul: Yeah, and so-
Heidi: I'm a parent, but mine are old.
Paul: And his point was, "Listen, parents have had it really hard."
And I was like, "I get it."
And so the reason I tell that story is, because I think that's a class of just interaction we're going to have to figure out as we open up, where I would say and where I think this is relevant for the work that I do is, what we're really talking about is individuals own risk assessments and own risk tolerance.
And there is an aspect that we don't talk about enough.
That's really purely, it's purely consent. It's the concept of consent.
Are you asking folks for consent to do the behavior that you're doing?
Is it appropriate to do that? Is it not?
And I don't think we often think about it that way, by the way, that wasn't even my insight.
Somebody used the word consent, that these are consent issues, and I was like, poof, "Oh, I got it, it's a consent issue."
Heidi: If only we had a framework around consent and contagious diseases that is like 30, 40 years old now, that would be super.
Paul: Well, it's funny, you mentioned that, because Alex Hidalgo and I were chatting on Twitter about this and he was like, "Masks are going to be forever now."
And I was like, "I don't think they are."
And by the way, I was very happy with the conversation.
People were chimming in, it was a very healthy conversation.
It wasn't a conversation that often the damn bird site, it goes to hell in a hand basket, in five tweets or less.
It was really good. And his point was like, "Listen, flu cases have been almost nonexistent."
I was like, "Yeah, I didn't get a cold this year."
And it's like, "Well, I get it. You're all wearing masks."
But I was like, "But, I just don't think we'll be able to do it as a country."
Heidi: Right. It'll be interesting to see if we do, and I think the effecting factors are going to be how much of the year is California going to be on fire?
Because a lot of the reason Asia is so good about mask wearing is they have so many high pollution days.
They do it for illness reasons, but they also do it, because the air sucks.
Paul: Well, yeah, and Kim, you were-- The day of Mars SF. Was Oakland also Mars, too?
Kim: I think the whole Bay area was-
Paul: So weird.
Kim: Yeah. I have friends in the South Bay, south of San Jose, that were seeing orange sky.
Paul: When we talk about the pandemic, and our conversation took a weird turn, it's very pandemic focused, which I'm fine with.
I don't know if people find this interesting, but I will say this, I handled the pandemic pretty well.
There were two times that broke me, like The Good Place, Chidi, like broke me, right?
Heidi: Now with marshmallows.
Paul: Right, exactly, I just watched that episode the other night.
That's exactly, you caught the reference, spot on, but I went to the grocery store and this was a time when they were limiting people in the grocery store, so you had to wait in line and "Okay, that's fine."
That was a little weird. But when I saw people starting to get into fights about who was where in the line that broke me, because I was like, "Social order is starting to break down."
You're seeing people fighting for resources in a kind of lizard brainy way.
And I don't mean that as insultingly, but just very needs and resources focused. So, that broke me.
And then Kim, I had a meeting at, I think 10 that morning, I woke up and I was like, "It's dark. Did I sleep? Is the clock? What the hell?"
I thought maybe the clocks were messed up, and it was five in the morning.
It was still dark. And I remember walking outside, and it broke my brain. It just broke my brain.
Heidi: So, I think that will affect mask wearing.
I think that the question of normalization is interesting and like, so I'm a sex educator.
I work with teenagers on consent based sexual education and to talk about how we've normalized conversations and protections over a generation.
I think that maybe we'll get to that with masks. Like, "This is my risk tolerance."
And that's like, "You don't get to tell me what my risk tolerance is."
Like, "I can't tell you what yours is, but like inbound, this is what I'm doing for myself."
Paul: I'll say it this way, kind of as a maybe good conclusion or kind of like summary, I think that's something as a country that we still struggle with, but I do like that this is causing us to maybe have those conversations in ways that are bringing kind of a new framing of things to the discussion.
I mean, I'll give you one other example, since you do sex ed work.
I was chatting with a bunch of my friends, and we were kind of laughing at all the parents who were figuring out the bubbles and the all that kind of stuff.
And we were laughing, because it was like, "Yeah, welcome to being gay in the 1980s," like you're serosorting, right?
Paul: And it's like-
Heidi: this is also familiar.
Paul: Right, the straight people don't know what they're doing, but like, "Welcome to the party."
And so, we were laughing about that.
I mean, not in an empathetic sort of way.
The thing is it's something we're still going to struggle with, but I also think that these conversations for those that are paying attention and want to learn more and pay more attention, they're getting a different view of a lot of these topics.