`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 1 of 71
`
`PX-170
`PX-170
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 2 of 71
`
`11/30
` return to office policy?
`s
`insider risk
`dr d
`/dei
`doing better on black+ , women , recruiting, retnetion, wrt google
`worse on latinx/native american
`
`news
`
`12/3
`ads- addtl exercise . j
`
` + p
`
`
`
`mitby: 1000 apps use 300 of them .... how to make that easy to find/access
`incremental
`moar deeplinks/appify
`
`reengagement / appify / deeplinking
`
`play pass - how to invest
`originals
`evolve
`
`v
`
`
`
`m
`
`12/9
`s
`
`? disappointed no spot bonus)
`declines (PITA pm - m w
`what is/is not payments data
`decline reasons
`messaging, changing billing dates ....
`-12% on starz
` - process for managing deps between us/payments
`
`
`0% is currently riot-starting in team, meme of impossible to police.
`either way, who is eligible is the crux. we dont have the kyc needed to a good job. dont
`even know if it'll be enough (apple associated accounts)
`we probably need IDV, background checks etc.
`hbo asia
`hotstar, wholly owned sub of disney
`
`constrained by SMB focus. e.g. large dev incentives can serve as discincentive for
`some of this bad behavior
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021362
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 3 of 71
`
`SQEX builds a spin off tomorrow and licenses some of their IP to this new company, will we fo
`to war?
`
`https://docs,google,com/document/d/15FiuVbrMiuU-aGHGJjfB7Cxm8A6irQHtZPz0-
`ZFFYvY/edit?resourcekey=0-xaao4FXiwuMBXUscs,Jl5bw
`
`comp
`
`-
`
`-
`
`comp plan - 3.6M
`450 base
`525 bonus
`2.6M equity
`cash flow: 3.3M
`
`announce, retroactive to jan1
`
`prioritize 99% don't pay full
`ok w janky rebate approach
`"savings to date"
`'independent' - ok w hbo asia
`? talk to developers .... get their instinct (sqenix example)?
`-
`try o draft off apple draconian practices, if any
`if they're applying, it could be really hard/angering small devs
`
`99% is the message that landed well, but will it sustain as program rolls out
`
`lock in something with finance so they're expecting it. might be matching apple, might be more,
`might be less (may or may not fly)
`make sure s
` aligned on plan and options and how we go to r
`IM)
`
`/p
`already communikcated it to r
`how do you want ot earmark, how might this affect optics of play plan?
`governance - pricing team needs to be involved
`c
` - how to get 'approval'/ set expectations
`
` (v easy for s
`
` to
`
`1. ok to delay an announce until we see how apple actually administers the program and
`what dev agitation may arise. {assuming there isn't burning pressure from prlgappldev
`agitation to do something to match apple]
`2. we should also do research with developers to understand how or whether they'd try
`to do the 'create shell company, license IP to it, launch new app' routine with apple.
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021363
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 4 of 71
`
`/p
`
`. s
`3. we need to set expectations with r
` has pinged them and planted
` to ensure they're aligned with
` and s
`the seed. we should work with c
`what (any) plan might look like (e.g. this is how much it would cost to match apple, this is
`what B and C might cost), and figure out how we 'earmark' the costs
`a. should consider if/how this affects the optics of the play plan
`b. governance: understand pricing team may need to be brought in
`4. other random s
` feedback/notes:
`a. would prioritize the '99% don't pay full 30%' message, over 'paying 0%'
`b. would be ok with a janky rebate approach, if it helps address T&S concerns.
`could have a 'rebate earned to-date' type thing in console
`c. need to get aligned on what devs would or would not be in the program. what
`ownership structure is ok, etc?
`
`d.
`
`payments team
`talked options
`converge
`issues: who owns, play sufficient control?
`
`status quo
`ngbf, but need needed to created deconstructed bf for us, because we
`only use some
`get out of buy flows altogether
`
`b ready
` - will go w shopping
`s
` s
`left w yt , ads, google store
`
`status quo works for us
`if yuo can deemphasize shopping or YT, put it into optimization
`processing, fops, risk,
`we could take web for digital
`ios flows
`ads, cloud
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021364
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 5 of 71
`
`-
`
`-
`
`to retain talent and have enough data
`everyone will go to apis
`- who wants to do ios
`to bundle
`
`deals
`rakuten sourcing/aggregating
`
`
` / p
`? escalation / c
`solve fragmented buy flows
`
`- will happen next week
`shopping - pis maintain api for 1 more yr.
`
`
`S
`- play pass t mo tuesday
`
`app squad wtf, camera?
`
`unity
`
`gpay- IDV
`clover
`s
` piracy
`app squad
`align w mmh on org and culture changes
`idfa CY
`deals prioritization
`
`org
`fix okrs
`
`20210105
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021365
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 6 of 71
`
`-
`
`DECISIONS/DONE
`dei okrs
`Advanced protection OK?
`half of play pass not in it.
`read up on potassium
`organize around gnosis
`https://gnosis.qoogleplex.com/p/ahNzfmdvb2dsZS5ib206Z25vc2IzchQLEgdQcm9qZWN
`0GICAgljhuYkKDA
`
`play leads
`2441
`
`2695
`s
`
`
`- what does potassium really mean? eliminate cookies, protect ads
`- what should my role be - resourcing?
`k
`
`
`exec presence
`ability to drive this big complex thing
`paradigm shift? from drive/comms to this?
`potential backfill for my team
`team fit, xfn influence
`-
`
`? pf
`android : headed to on-device measurement/ tracking
`
`3048
`
`p
`
`pb
`
`
` feedback .... how to operate
`s
`kyc / developers
`indians, covid, job loss
`loan apps shady, aggressive collections.
`needed RBI license
`hard to see APR due to escalation
`
`supercell, 1 QOM -> 15%
`apple ok w sku parity ( we could enforce via gvp)
`runway
`raising money eom jan.
`FTjan
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021366
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 7 of 71
`
`a
`d
`
`-
`
`HTfeb
`start masterclass 1/march
`store - autopilot for 6 monts ....
`new ranking for games
`deals, mekka, he
` - good place, stepped up
` needs to build relatinoshi0ps
`y
`
`pf
`gmscore
`pm/eng issues
`d
`
`runway
`. .. thinking abt the small devs
`/k
`needed more a
`mmh - thought they'd be too easy on devs
`implementation mtg
`india mtg
`EU legis
`?
`a
`get someone for a
`free 1-200M
`post-install x-sell, more ads, less organic
`search page, depth ads
`deals
`
`, 1 person for k
`
`/m
`
`
`
`-
`
`-
`-
`
`-
`
`msft
`tablets
`
`b*
`
`more ammo to care about large screen
`
`6024
`s
`
`/play leads
`'very good' for play .... ~350, without XFN, without extend
`plan for 80% (at lower end)
`some may be granted late in year, details
`- YT
`
`
` / s
`h
`'so much work', nothing of substance. talked to s
`
`.
`
`20210106
`DECISIONS/DONE
`CSM -> m
`
`, privacy
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021367
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 8 of 71
`
` on pb sitch
`briefed a
`saw rubidium, t&s updates
`
`devexp
`privacy
`kyc
`
`rubidium
`who is the director to rally and focus
`
`s
`
`
`org/culture
`reorg, people coming loose.
`
`pb
`1 00M cushion
`app squad 1 P focus
`fitbit sub lead
`he
`
`350 to play, epuc/bd,
`caveats
`45 - allocated to extend location
`25 - give back to d
`. can probably fund the rest of the 60 himself
`20% holdback release in october
`rate card, can trade heads for location
`complexities
`negatives we've been carrying
`gmscore - slated for 22, under the extend
`unicorn - j
`
`dont know what happened to yt heads
`
`20210107
`
`runway
`rubidium
`clover
`EU new deal
`unicorn
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021368
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 9 of 71
`
`k
`
`
`
`exec presence
`ability to drive this big complex thing
`paradigm shift? from drive/comms to this?
`potential backfill for my team
`team fit, xfn influence
`
`20210111
`DECISIONS/DONE
`review hwgpw
`advised pk
`updated leads on stuffs
`
`xfn
`
`pf, AM conscripted into dsa/dma
`dsa - moderation
`dma - p2b 2.0
`new deal - consumer protection
`
`cultural shift doc
`
`PK 1:1
`
`ios squad
`PM, mktg, comms, 'central team'
`? working w m 's team?
`uer to figure out why not engaging w G products . e.g. security w/faceid, multi app use
`cases
`? how are 1 P teams orged wrt ios vs android
`much smaller?
`teams were too NIH. wouldn't use apple's libs
`much more constrained : easier to dev for .... make that 1 way better, rather than
`giving ppl 10 ways to do it
`
`contributing some teams
`search - well staffed, corralling
`tasks - 1 ios dev, get any help they can get
`apple - only 1 enterprise acct
`
`10 ppl CT - 8 eng 1 ux 1 pm
`trying to id people who would work in a platform capacity
`comms, pr, mktg
`catch product teams output
`needed incentives, company level OKR
`still no exec level view of what we want to do on the platform, no exec sponsor
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021369
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 10 of 71
`
`e.g. dealing w privacy nutrition label
`
`20210112
`TODO
`
`s
`
` I hgpw
`
`app quality
`pixel ux
`ios generally more polished
`more effortless across form factors
`
`a
`
`
`
`not as forward as m, gh
`shadow of pb
`
`
`jalwa 15 for line 10
`some given away to wear, maybe.
`stadia - fertile recruiting
`j
`
`
`-
`
`-
`
`- worried we need full time person
`p
` nog good at brining people together and getting them over hump
`
`s
`
`m
`
`-
`
`
`buyflow
` - getting beat up by bill ready
`k
`/c
`/b (?) -> p
` to make call
`v
`b : going to force all physical goods stuff to their other buyflow
`3 options
`support deconstructed only
`b should handle all physical, play should handle all digital
`-
`g1?
`single consolidated flow across all of google
`gpay logo. caesar told stop. thought it was only india. we thought global
`restitution ph 1 - current, ph 2 - new cases. ph2 not on track. kftc pressure
`trade off br comp vs restitution
`not enough vis to understand tradeoffs
`
`-
`
`-
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021370
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 11 of 71
`
`20210115
`h
`
`play pass funding
`gc, fop ownership. XXX doug
`big nums, 1 M ppass. subs 1 OOM
`
`42eng years disingenuous, and play would have to take that cost and put billions at risk
`
`
`
`s
`d
`
`-
`
`-
`
`-
`-
`
`
`stress for android
`c
` can't deal with him
`creating some issues out of non issues. or adding complexity
` too quickly .... is there some fomo?
`XXX name drops s
`/escalate to s
`/t
`bring to s
` first
`- wasting time talking him down
`lot of shit talking .... shit talk abt s
` to c
`you 4 run incremental
`
`, r
`d
`, l
`, c
` disengaged.
`felt microed, so s
`/a
`gmscore/play sync - 20 ppl in room.
`ed proposal for 3rd party stores
`
`/d
`ejc does not fuck around. told to talk to a
`d
`: ejc not solving problem. throwing ppl under bus
`
`.
`
`. shit talk l
`
` to c
`
`v
`
` going to leave
`c
`: d
`
` rubbing off on v
`
`
`
`mmh
`
`pm's having trouble w m
`
`?
`
`https://docs.google.com/document/d/1O imfkvktVnOUYZGm4Y5sKkiiTy2L bx vekfxt eYU/edit
`?ts=6001 e875#heading=h.3uiis2gsnri n
`
`20210119
`-
`thx for making time
`huge thx to teh teams working behind scenes to distill into 10 slides.
`put landability aside
`focusing on principles first
`
`-
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021371
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 12 of 71
`
`legal advice
`discussion is complicated and easy to get derailed. we had a structure in mind so we're
`going to try to adhere to that.
`only 10 slides of dense info
`
`20210120
`inauguration day
`
`/s
`
`
`
` d
`
` et al
`see issues first hand
`strat roadshow w s
`cros update
`
`s
`
`tablet
`
`he
`
`XXX DEBUG OHANA , DUMBLEDORE, B* ASK
`
`s
`
` +9
`
`perpetually positive
`
`runway/sr leadership
`transparency as principle
`CN app store approach ... 50% for discovery etc
`s
`
`existential issue: will this stop anyone from complaining
`need to have billing optionality
`like more transparency. what is the way this would be explained to world
`integration requirements stay? or is everyone in?
`making people pay that weren't paying
`do we need more enemies? only 125?
`misaligns us w devs
`dev fee can be levied
`hobby developer license
`tiers of support
`very google - rational, cerebral, legislators/regulators need bitesized
`takeaway
`
`-
`
`j
`
`
`
`'s points -- is that what we believe about the
`the one comment I'd add -- to S
`likelihood of being required to offer choice of billing system seems fundamental to the
`direction we choose here. I hadn't heard until today that we believe that;s a highly likely
`outcome.
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021372
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 13 of 71
`
`d
`
`
`
`strategy
`runway
`clover
`
`
`spotify/s
`new agreement, might cost more
`not necessarily more commit
`tk wants more$, GM negative
`s
`: existing deal may have to open
`not just more cloud spend to save $ on play?
`
`store process
`2 teams own
`walk up chain to get guidance
`working on different goals - ipd / interstitial rate
`not enough escalations
`
`clover/s
`
`
`status ... b*
`runway/s
`
`? bulk
`step back, pressures, kr,
`need perspective / game plan on whether or not proactively before apple is a good idea,
`waiting till regulatory/bitter end, know what we want/particular event
`gametheory
`shopping -angry, defend, instinct to fight remedies were lame (not really solving the
`problem, trying to give the least). until clear it wasn't work. not clear what counterfactual
`is
`nothing yet to make majority to stop agitation
`dynamic w apple .... divert firepower, but is the collateral damage going to hurt us?
`: was it a menu ..... make it a recommendation by doing scenario planning
`we're focused on regulatory ... but what about competitive .... if we hold line too
`long
`
`j
`
`if you push tencent, ramping their store, ilke etc, join up w samsung
`if we did everything tmw, what would epic's response be, supercell etc
`rate card - impact on ecosystem, who gets riled up, new enemies
`epic going to continue, ags, dma
`choice of billing will continue as a vector
`20% narrative
`sound bite
`ags/epic/eu - meh?
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021373
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 14 of 71
`
`is it worthwhile doing up front
`how to integrate j
`'s team. without pissing current team off.
`
`samsung/wear
`wear addendum in samsung deal
`wearables app
`merchandizing for watch apps and watch faces from galaxy store
`categories
`entertainment carousel 20 curated
`need apis from play
`m
`
`
`need public api
`'can be curated' -tooling to populate the carousel
`
`/cl
`emu
`storex
`better together
`commercials
`
`
`
`j
`
`s
`
`pb
`
`pk
`
` + m
` -ATC betw h
`t
`debug 2 pager games vision. w g
`d
` - s
`
`pms under him being too territorial/dictatorial
`
`
`
`h
`
`
`
`OKR review
`'my business requirement'
`authoritative influence approach. doesn't want to convince engineers
`
`incremental review w s
`GTM fighting over 50 vs 500
`distribution - not allowed to say no
`10-ish
`
`
`
`focus on regulators - defend or make peace ?
`you got a business idea, we have a business model for you. we want you to be successful
`dont simplify, be comprehensive
`stores - as a nother dimension
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-00702137 4
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 15 of 71
`
`s
`
`
`
`play store for wear
`
`AGG review
`? how to for load times. journey for debugging perf, etc
`
` passdowns
`
`s
`
`s
`privacy
`nutrition labels, ATT
`thoughtful approach, but apple more aggressive
`
`better together importance
`e . bluetooth, everything that would be common in all the better together scenarios
`
`pf
`
`risk he
`compliance he ask - heard it was granted, butl
`
` F claims not to have received.
`
`go/play-loyalty-datasite
`
`d
`
`
`
`- a
`
` took on scope and then ...
`
`runway
`- not clear billiing option viable
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021375
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 16 of 71
`
`FEB
`
`a m
`- over index on impact/numerical impact.
`focus on KS As.
`-
`- bias to execution rather than strategy
`
`mainline
`- android: op pain. not staffed for operations pain
`first place to collab, release tooling & operations
`-
`- graduating code from gcore to framework
`- mainline focus on aosp
`- gcore focus on 1p dev velocity
`- what to combine
` adversarial on hc .. .'then we'll have to ask'
`- d
`- a
` speaking on behalf of london? london annoyed at s
`
`?
`
`
`s
`passdowns
`- s
`
`- dramatically improve privacy narrative
`- big narrative at io
`- potassium, assistant, privacy nutrition labels
`robinhood, how stuff rolls out/fans out/butterfly effect. if not
`-
`careful, always in any cycle related to online svcs
`
`- better together, phone + watch
`- dei
`- m joining s
`feb 9 - s
`-
`
`'s team
` updating on silk
`
`/g
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021376
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 17 of 71
`
`-
`
`feb 16th - roundtable type thing
`
`wear - slsi, faster than tizen, smooth,
`
`k
`
`-
`
`
`redone all content in store, all formats, all surfaces, just about
`everything that's not ranking
`- gets more value out of existing real estate
`- created app hero card, used less pixels, more effective
`
`alignment
`- 65 execs, aligning on company vision and okrs
`- 5 days. 2.5 days was actually focused on mindfulness
`- mindfulness
`-
`listening, being empty
`language
`trust
`-
`
`rackets
`
`-
`-
`
`n
`
`
`- collab/perspectives
`
`2/8
`
`s
`
`
` not aligned with lower levels
`/c
`- s
`-
`'asks' vs collab problem solving
`- how to work as one team
`trust has been hurt
`
`-
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021377
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 18 of 71
`
`- aiai, sandboxed app, hotword processing
`- android told aiai to move stuff out of gmscore
`feel like trying tobe played
`-
`
`2/9
`passdowns
`- ~60 mins assistant review
` last thursday
`/r
`/h
`- 60 mins pixel review w s
` said @ gleads .... recap, v well done, excited, great review
`- s
`- pixel android 2021
`- what are the spaces on the mobile phone going forward
`- assistant
`- voice strategy
`- gboard dictation ++++++ responsiveness, grammar,
`punctuation, gmail, warm words (stopping alarms, from
`lcokscreen etc)
`- smartspace
`- at a glance (next mtg etc)
`- positioning as what assistant thinks you should be doing
`right now .... postit right in front of you
`- at starbucks - loyalty card, at airport - boarding pass etc
`- silk/material next
`leak. partner. out of date, inflight
`-
`- silk - make android modern & reliable. color themeing, ssytem
`spaces.
`- visual ssytem , IA update to all spaces
`- everything is getting an update
`- material next
`- systematicization of that
`- more app side
`fewer jumpcuts
`-
`- 4 core spaces
`- quick access
`- attention
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021378
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 19 of 71
`
`- smartspace vs notifications. help you based on
`context. extension of notifications
`- yours
`- home screen. your mess
`- everything
`- all apps. search. sortable/filterable. when in doubt
`go here. 1 index
`
`- coarse gestures
`
`2/10
`s
`
`-
`
`
`feedback on the ux session
`- headed in right direction
`-
`just need next level of detail to expose thorny issues
`- how will teams really make decision of what makes it onto the ux
`- will consumers understand it
`-
`runtime dialgos turnin ginto tos
`- p
`-
`family review. m
`- b
`g
`
`/p
`, j
`
`, j
`
`/mmh/t
`, ss
`
`
`/r
`, apps team council. accepted
`
`- material next challenges, sufficiently plugged in?
`-
`read in on jetpack
`
`- clover
`-
`'no penalty', defensive move. THERE IS A PENAL TY
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021379
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 20 of 71
`
`j
`
`r
`- GTM dates/reasons, branding
`- need to do b* and clover
`- he
`- yeah key points are (1) we need a hedge, and (2) we need TAM and can
`converge later
`- HC is still an issue, but if we think some form of "fund out of ATL" or
`other bucket, then it's a credible option.
`
`- gmscore
`1) there is an overhead of this new platform, with new unique
`-
`constraints (e.g. adding production monitoring, alerting, custom
`rollouts) to account for the new platform. given how different it is, we
`need to do something here for that or suffer samsung S21-esque bugs
`(also similar for mainline/ "google play system updates")
`-
`- within gms there are two complexities ...
`-
`2) the APls in GMS presume a certain security model, which doesn't
`exist on any of our short terms solutions (is orthogonal to hyperv vs
`haxm). basics like not storing PII on device.
`- we need to do a full security audit of all APls w.r.t the short term
`security model and see how we could either modify (and accept risk)
`or disable apis
`- and implications of that on app catalog
`- 3) there is a massive amount of logistical work to ensure all teams who
`are building are qualifying their work on/ considering the platform
`the long term solution with microsoft simplfiies this for us because we
`can provide much stronger security guarantees
`- did that make sense?
`- so to answer your q ... it's not really a win10/win11 thing. its a "battlestar-
`esque security model" vs our nested virt security model
`- nested virt requires msft to rehost windows on top of hyper-v
`(effectively depriviliging it)
`
`-
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021380
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 21 of 71
`
`-
`-
`-
`
`so it's a shitload of work for them
`(or equivalent other solutions)
`they are aligned that it's the right long term direction, but obv is a
`massive amount of work and is not feasible for their fall timeline
`
`---- CLOVER HC -
`T
` L , Yesterday 2:33 PM
`re:hc and doing 8* and clover ... .i think we need to more clearly articulate, in
`the b* case, how much more he needed, for how long, and why (TAM +
`tablet strategic value+ revenue)
`for clover, assuming we scrambled all the jets onto it, i'm unclear as to what
`the resourcing gaps are (sorry i don't have the latest decks if they're there),
`and for which teams (e.g. i'm pretty sure android is a big hole), and what the
`sequencing for MVP launch would look like
`*resource sequencing for MVP launch
`
`, Yesterday 2:35 PM
` M
`A
`What's the right forum to get signal on clover funding even without the
`backwards compatibility piece?
`
`T
`j
`
` L , Yesterday 2:35 PM
` is going to raise the issue w s
`
` tmw morning
`
`, Yesterday 2:35 PM
`
` M
`A
`Sorry lag.
`Ok sounds good
`
` L , Yesterday 2:35 PM
`T
`but not sure he knows what the number is, or where the holes are (play vs
`android)
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021381
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 22 of 71
`
`, Yesterday 2:35 PM
` M
`A
`For what needs funding on clover. We have imprecise ballparks
`Would be helpful to get an envelope where we can apply heads and hold
`some reserve to cover holes as we see them. I think our original estimates
`are ballpark accurate
`
`T L , Yesterday 2:37 PM
`(can you point me at those ballparks, i assume it was in last deck to h
`somehwere, but i don't have that handy)
`
`
`
`, Yesterday 2:39 PM
` M
`A
`It was ballpark 50 play+ gms, 50 android os
`The 100 is about right in total but the balance may shift
`
`T L , Yesterday 2:40 PM
`sorry is that assuming we killed b* and abandoned win10?
`
`, Yesterday 2:40 PM
` M
`A
`(we had other asks for better together features, developer tools,
`asap/policy, etc)
`
`T L , Yesterday 2:40 PM
`and what level of security model you think is implemented by october?
`
` M
`
`, Yesterday 2:40 PM
`
`A
`Yes.
`This assumes no back compat
`I think the backcompat thing in parallel would be 30 ballpark on top of that.
`The old estimates aren't probably accurate enough because they were high
`level
`Can put something together with more detail now that we've had convos
`with them
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021382
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 23 of 71
`
`T L , Yesterday 2:41 PM
`and you think 100 would make apps avail by october? (ie full android security
`model)
`
`, Yesterday 2:41 PM
` M
`A
`Mainly would need to know what level of detail you are looking for
`There is no path for us to have all apps available by October
`That requires joint work on both sides that neither can commit to
`
`T L , Yesterday 2:42 PM
`what level of security model would we accomplish by oct?
`
`, Yesterday 2:43 PM
` M
`A
`We can have a beta by October on win 11 (hand picked apps + games), and
`scale up to the "long term" solution with microsoft
`It would functionally be Battlestar
`
`T L , Yesterday 2:43 PM
`is the limiting factor the android security model, or gms compat?
`
`, Yesterday 2:43 PM
` M
`A
`In terms of security model it's roughly identical. The large Delta in scope is
`that we would need to figure out how to onboard Google Play services, take
`on deeper os integrations, and rearchitrct a bunch of stuff
`It's a combination. The Android security model is a completely made up
`dependency so we could just arbitrarily decide to ignore it
`The Google Play services one will be quite tough to get around. It's just a lot
`of work + risk
`(also is tied to the same mental hurdles we would have to cross to relax our
`security requirements)
`
`T L , Yesterday 2:45 PM
`the gmscore stuff is orthogonal to win 10/ win 11 right?
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021383
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 24 of 71
`
`i'm assuming that choosing to rest andorid on hyperv impacts android, but
`not gmscore .... pls correct if i'm misunderstanding
`
`, Yesterday 2:46 PM
` M
`A
`I can at least take a closer stab at scoping out the work if it helps, but given
`how much scope we've uncovered in our convos
`Sec getting to comp
`GMS is more about security model and deployment model
`1) there is an overhead of this new platform, with new unique constraints
`(e.g. adding production monitoring, alerting, custom rollouts) to account for
`the new platform. given how different it is, we need to do something here
`for that or suffer samsung S21-esque bugs
`(also similar for mainline/ "google play system updates")
`within gms there are two complexities ...
`2) the APls in GMS presume a certain security model, which doesn't exist on
`any of our short terms solutions (is orthogonal to hyperv vs haxm). basics
`like not storing Pl I on device.
`we need to do a full security audit of all APls w.r.t the short term security
`model and see how we could either modify (and accept risk) or disable apis
`and implications of that on app catalog
`3) there is a massive amount of logistical work to ensure all teams who are
`building are qualifying their work on/ considering the platform
`the long term solution with microsoft simplfiies this for us because we can
`provide much stronger security guarantees
`did that make sense?
`so to answer your q ... it's not really a win10/win11 thing. its a "battlestar-esque
`security model" vs our nested virt security model
`nested virt requires msft to rehost windows on top of hyper-v (effectively
`depriviliging it)
`so it's a shitload of work for them
`(or equivalent other solutions)
`they are aligned that it's the right long term direction, but obv is a massive
`amount of work and is not feasible for their fall timeline
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021384
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 25 of 71
`
`T L , Yesterday 2:53 PM
`for my education - the b* security model - did we actually disable gmscore
`apis or modify them specifically for b*?
`
`, Yesterday 2:53 PM
` M
`A
`another basic example that came up that shows why things balloon in scope
`we disabled them
`we audited all games
`looked at what apis they used
`and then disabled the rest
`(games notoriously have few gcore dependencies)
`the reason i am advocating for an envelope approach is that if we're serious
`about the project, we should make sure we have reserves to handle massive
`bogeys
`an example that came up yesterday --
`the memory management model for windows and android is dramatically
`different
`android is greedy
`so what happens in a case where say
`you're using signal or what's app on your device
`on your windows pc)
`you close it, and then a notification comes in
`either (without massive surgery) we have the VM running in the background
`causing large overhead on the windows device, despite you having close all
`android related stuff
`or we disable things (background services don't run)
`or we need to do some surgery
`things like that
`
`T L , Yesterday 2:55 PM
`or you install the signal desktop app □
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021385
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 26 of 71
`
` M
`
`, Yesterday 2:55 PM
`
`A
`ha ha
`that example applies to all apps that use background services
`~most apps in the catalog
`□
`just illustrating the point
`
`T L , Yesterday 2:56 PM
`yeah just being snarky
`
`, Yesterday 2:57 PM
` M
`A
`i can enumerate a bunch of these if it helps bolster
`
`T L , Yesterday 2:58 PM
`(trying to decide what it's worth to try to overload my brain on this)
`
` M
`
`, Yesterday 2:59 PM
`
`A
`lol
`
`T L , Yesterday 2:59 PM
`(as i listen to rubidium mtg)
`
`, Yesterday 2:59 PM
` M
`A
`tldr, there are a bunch of examples like what i mentioned
`ha ha
`in general the thing i'm worried about is that we kind of need a blank check
`on this
`with some sense of what that upper bound is
`i'd advocate for an approach like t his
`we tell s
` / h
` this is ballpark what we need ceiling
`we can come to advocate for 50hc blocks to unlock
`or some such arrangement
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021386
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 27 of 71
`
` L , Yesterday 3:00 PM
`T
`i get it.. . .i dont think its realistic
`scope is a massive lever for october
`if all they care about is having tiktok, insta running by october as a beta, that
`releases a lot of pressure on trying to find perfect (or even good) solutions
`if you knew that for a fact, would/could you scale the ask down? (obvs i
`wouldn't start negotiations there)
`
` M
`A
`(sec brb)
`
`, Yesterday 3:07 PM
`
` M
`
`, Yesterday 3:16 PM
`
`A
`back
`that would be predicated on a few things
`1) we'd need the developer to be willing, and to want to sign up knowing the
`reduced security (would be a big deal for FB i assume -- especially allowing
`FB login)
`2) (also non-technical) we'd need to get the android security team on board
`with it, they've historically been more challenging
`3) a lot of the scope is coming from msft asking us to do integrations with
`their stuff. unclear if we could negotiate them down off of that
`(also relates to #4 which is a lot of the scope is around modifying system ui,
`and constructs like notifications etc in android to adapt to windows
`standards and feel native)
`the current october scoping was always trying to find a passable solution
`that is tailored to a set of partners we hand select
`we were planning on being even more conservative than what you
`mentioned ... by identifying a set of apps that don't use very many gmscore
`dependencies ... aren't security sensitive ... etc
`and starting from that angle rather anchoring to a list microsoft wants. they
`seemed OK with that approach but we haven't spoken in detail. also we
`didn't tip our hand that GMSCore is not supported in b* ... we couched the
`whole discussion around security and left it at that
`
`HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
`
`GOOG-PLAY-007021387
`
`
`
`Case 3:20-cv-05671-JD Document 387-2 Filed 02/09/23 Page 28 of 71
`
` Yesterday 5:32 PM
` H
`G
`my sense is we should probably write down a "what does both look like" and
`call out where there's efficiencies between B* (shared components) vs. the
`delta the Microsoft creates for Clover. then meet to discuss, recognizing
`that 50+ heads is unlikely to go anywhere, might be worth pushing on
`assumptions and see if the exercise yields somethingn aroudn 30-40 HC in
`2021. is that a good next step? other thoughts?
`(bottom line, I'm trying to avoid the Clover