Meta & Children's Online SafetyChat: imagine going out in public and telling people we rank their kid's facebook by what makes them consume the most content IDs lol
Thread View
████ name redacted. Different bars may be the same person or different people.
quoted in an excerpt in the archive.
Friday 26 January 2024
13:25████03.11.25 's qual would have been a lot more impactful if she had a better understanding of the platform, the youth prev reduction workstream, and the central decisions that the work could inform. We need to make sure that all of our research is oriented in what is actually happening at the company. It's super inefficient when we don't. Not useless, but not maximizing our impact.
13:27████Stuff like MYST is so far away from the product teams it is destined to be tangential to product impact, you need to reinforce to the team that it is their job to know what is happening on our platforms and across the company.
13:28████and need to build relationships on XI and FI
13:29████and honestly, on SI. There is still no meaningful engagement from YWB -> SI-X, and it's a huge gap in their understanding.
13:30████reached out on baseline protections because she understands this, so props to her. She has grown a lot in the last 18 months.
13:32████why are and are not tying the threads?
13:33████because that's not what the PM function does
13:33████they need to though
13:33████no, WE need to
13:33████they cannot operate in their little microverses
13:33████tell your reports!
13:33████oh yeah we need to as well
13:33████i need to
13:33████I am!
13:33████not today. today is HR freak out day for me
13:33████dealing with bunch of nonsense fml
13:37████slightly misspoke about an appearance comparison finding yesterday, and tried to correct but was not clear enough... and it seems to have created a minor issue. I'm not telling them yet, I'll just sort it out because it's not worth their time, but I need you to coach people on how to present data efficiently and very accurately
13:37████you and are strongest in data communication in the team
13:38████what'd she say
13:38████and why is it an HR problem
13:38████oh that is not the HR problem
13:38████ahh
13:38████what'd she say
13:38████HR problem is something i can't share
13:38████she said teens are getting exposed to apperance comparison content almost exclusively through recommendations vs organic
13:39████lol
13:39████that's not slightly misspeaking
13:39████what a strange thing to say
13:39████what she meant to say was when people provided examples of apperance comparison content, the examples came from recommendations vs organic
13:39████because they had that chart in the appendix
13:39████oh she was describing her own research?
13:39████yes
13:39████the diary study
13:40████ehh I mean within context that's not so bad
13:40████which cannot state that
13:40████if she'd just out and out said it that would've been very bad
13:40████no it's bad because people automatically interpreted it as prevalence
13:40████yes but there is presumptive context when presenting a deck
13:40████and then was like "no no, this is not prevalence" but in 43290 words
13:41████and then was like "yeah because this is qualitative data and not generalizable"...
13:48████kk
13:48████yeah they should work on that
13:48████has always struggled with concision, is usually pretty good on this bad that's a miss
13:49████but*
13:49████who was the audience?
13:49████I would assume any reasonable product folk would be able to navigate that just fine
13:50████CSI leads
13:50████etc etc
13:50████someone from policy
13:50████was there.
13:50████ehh i'm not enthusiastic about this pathway, gotta stop using ICs for pet projects
13:50████too I think.
13:50████they haven't even presented to FI PM
13:50████they signed up themselves!
13:50████they see the incentives on the wall
13:51████it's crazy to me that they're not presenting to the team that is midflight in deciding how to treat this content
13:51████to be fair, and read the deck end to end and it's getting traction
13:51████classic central team silo
13:51████yeah, they should.
13:51████before they talk to FI we need to make sure they are not misspeaking about the data though
13:52████ehh, FI would be fine because they're product people
13:53████is harder, need to spoonfeed them
13:53████FI PM would be like "do you mean this or this" until it's absolutely clear
13:53████every time I present to them they grill me on specifics, it's great
13:54████half the time that's where my next couple weeks go following up on useful stuff, whereas you see the SI- X leads comments on s deck and it's like... randomizing nonsense
13:56████the FI collab on youth is going very strongly, need to get YWB to align with what our org is actually doing that is impactful
13:56████Shared an image
13:56████yeah, of course a more analytically literal xfn group would be fine-r but we still need to be very accurate in presenting data.
13:56████and concise
13:56████no room for misinterpretation
13:56████I would rather land impact than take no risks
13:56████FI PM would grill and understand.
would go and tell I'm incompetent
13:57████I know disagrees because he said it to me directly
13:57████is gone because nobody cares what she thinks
13:57████is the lead now, and is a close ally
13:57████I want to take smart risks.
So i rather coach them on conciseness
13:57████also because it's my job
13:57████It is much less important that they land impact
13:58████it is much more important that they orient themselves to land impact
13:58████these niggly little things and giant caveat slides are just background noise
13:58████just depried crisis for UXR. tooth and nail.
13:59████(Y) it's a start, we need to de-pri it for SI-X, but it's part of portfolio and he cares more about that than what makes sense as org strategy
13:59████I really don't care that team is super passionate about this. i'm not allocating half of on something where there is 1 engineer.
's time
13:59████(Y)
13:59████needs a focus area
13:59████misinfo is free
14:00████she is too horizontal atm, no path to 6 that way
14:00████I'm thinking 50% misinfo, 50% backfill
14:00████ehh was doing fairly random stuff
14:00████There is a neat product efficacy measurement project should do. and maybe also help with floor of protections
14:00████floor of protections is random enough as it is
14:01████maybe also look at problematic use thresholds
14:01████what is product efficacy measurement, just evaluating things?
14:01████PU is pretty dead in the water
14:01████isn't already on it?
14:01████needs to be out of it she is burnt
14:01████fair
14:01████for PU the question is when / if it is effective and how we trigger
14:01████it's not effective, it's not meant to be
14:01████it's a PR product and that's fine but shouldn't be staffed
14:01████efficacy: does anything we do help... in anything.
14:02████too broad
14:02████I'm starting to think the same tbh
14:02████I'm over problematic use, just that it's in every single legislation
14:02████start with the outcome we want to move, evaluating products only makes sense if there's an actual implicit goal
14:02████the legislation says stop optimizing for vpvs
14:03████yah
14:03████that's where the juicy work is
14:03████what is an alternative to our current toplines
14:04████in a review we were in last week said she needed research to see what features are optimizing for time spent. i wanted to bury myself somewhere.
14:04████it's like.... yes.
14:04████collating org goals is a good project
14:04████what do you mean?
14:04████tell me more
14:04████get each surface team's goal outcome that they're optimizing for, put in doc
14:04████like GYSJ optimizes for group joins, feed relevance optimizes for vpvs, IG feed optimizes for sessions
14:05████gotta meet now. but is that not a
14:05████to my knowledge this is not consolidated
14:05████does do projects?
project?
14:05████idk does he not?
14:05████he could do it, but someone has to
14:05████it's not hard, but it is important
14:05████good starting point for "what do we optimize for; is it defensible to do so for teens; what should we optimize for"
14:06████came up in the product policy VP / XI discussion around youth, is probably the one pushing it to diana
16:10████seems more promising than whatever roadmap is... they simply cannot do another half of finding themselves
16:10████this would be the what, 4th in a row?
16:29████4th year
16:37████They should audit every surface team's goal across the company, think about whether we would be comfortable telling the world that we are optimizing for x for their kid, and when the answer is no, come up with a defensible alternative.
16:37████Do they still have too many PD?
16:38████Too many and they can't work with ig without fighting
16:38████that's fine, keep them busy fighting IG while they audit the company's goals
16:38████imagine going out in public and telling people we rank their kid's facebook by what makes them consume the most content IDS lol
16:39████this will be a public conversation and we are not ready for it
16:39████and then when they hear what things keep their kid consuming the most content IDs sorting their feed by controversial, eye-popping filth we're off to the races on a really fun conversation if
16:40████is on paper supportive of this, we should seize the day and fund it
16:40████She is But she is also quite uninformed
16:40████I'm not ready for the weekend
16:40████then a click deeper, do the same with each value model term... like should we rank your kid's feed by.... p(like)-yes; p(most comments read)-no
16:41████this would be a really cool research project
16:41████Well1.... Maybe yes on p most comments read if the comments are decent
16:41████they never are
16:41████comments are the worst part of facebook and everyone knows it
16:42████yet that's what we're optimizing for because people can't help but watch people fighting
16:42████IG?
16:42████it's the same thing with the car crash videos kids are seeing, humans naturally look
16:42████IG comments are also awful
16:42████because their UI is asymmetrical
16:42████the negative feedback option on IG comments is genuinely invisible
16:42████Yeah
16:45████AA this is a much stronger path to impact than anything else on
16:45████make someone do it
16:45████████?
16:45████she could do it
16:46████depends if there are serious measurement needs or not
16:46████s roadmap ok I'm brain dead