Merge pull request #76 from jorydotcom/minutes-2019-10-08

Adds minutes for 2019-10-08 Face to Face meeting

meetings/2019-10-08.md

@@ -0,0 +1,279 @@
+# **AC New York F2F Meeting 2019-10-08**
+
+Agenda: [https://github.com/ampproject/meta-ac/issues/47](https://github.com/ampproject/meta-ac/issues/53)
+
+Facilitator: [@tobie][tobie]
+
+
+## **Attendees:**
+
+*   Elisa Budelli - [@elibud][elibud]
+*   Joe Alicata - [@wirelessjoe][wirelessjoe]
+*   Levi Durfee - [@levidurfee][levidurfee]
+*   Maggie Wettergreen - [@mjwettergreen][mjwettergreen]
+*   Pablo Delgado - [@pdelgadorodriguez][pdelgadorodriguez]
+*   Sumantro Das - [@sumodas][sumodas]
+*   Ted Shuter - [@TedShuter][TedShuter]
+*   Tobie Langel - [@tobie][tobie]
+*   Kenji Baheux - [@kenjibaheux][kenjibaheux]
+*   Tim Jones - [@tones][tones]
+
+## **Excused:**
+
+*   Melanie Sumner - [@melsumner][melsumner]
+*   Melissa DePuydt - [@msteffan][msteffan]
+*   Senthil Padmanabhan - [@senthilp][senthilp]
+*   Charles Vazac - [@cvazac][cvazac]
+*   Guilherme Moser de Souza - [@mobtec][mobtec]
+*   Jervay Singh - [@jervay][jervay]
+*   David Merrell - [@dymerrell][dymerrell]
+*   Ali Ghassemi - [@aghassemi][aghassemi]
+*   Terence Eden - [@edent][edent]
+*   Léonie Watson - [@LJWatson][LJWatson]
+
+# **Minutes**
+
+Resolution: delay meeting by 30 minutes so Kenji can make it.
+
+Potential topics:
+
+*   Process, roles
+*   Review project board and open issues
+*   Foundation
+*   Afternoon: ESCAPE workshop, SXG + web packaging
+*   TSC: talk about how the AC can give effective input to TSC for steering the project in a given direction.
+
+
+## Process
+
+Scribe: [@kenjibaheux][kenjibaheux]
+
+Levi: organizing this is a lot of work, Tobie is doing too much. We could help.
+
+Tobie: comes with the responsibilities
+
+Elisa: if we could have a private slack, it would make things easier.
+
+Tobie: we have a mailing list, we have a slack.
+
+Group: slack is hard to work with for a couple of reasons, not part of everyone’s normal work loop. Email would work.
+
+Tobie: group was originally seeded via partnerships. Some self nominated, some reached out. It does result in situations where folks are not necessarily close enough to the problems, or may  not have the time or motivation to participate, which is normal. Hence, the idea of renewing membership based on participation, contributions. Rather than involvement, what matters is the output being created. Proposing to focus on outcomes rather than presence (nods).
+
+Maggie: do we have a task management tool? Due dates?
+
+Tobie: github tooling, part of the problem is that we want to have everything in the open.
+
+Pablo: do we have the time  and manpower to create outcomes?
+
+Sumo: advisory committee’s role is to steer the direction of the project, give input to TSC. How do we communicate prioritization queues? 
+
+Kenji: I took an AI on following-up with the “give us access to the user research”. I did my best but this was really through internal threads and so on which is not something that everyone  can do.
+
+Tobie: the closest model I have is the w3c TAG group. Their role is to make sure that the requirements of different stakeholders are integrated, think about the big picture, setup processes and getting involved. They write findings, high level opinions  about things. One example for us is around a11y: we’ve noted issues, discussed them, and  reached out to TSC/... Alternative is through back channels but...
+
+Kenji: downsides of back channels, it could be one-offs, it has to be part of the process.
+
+Sumo: 1 pagers to not lose the audience
+
+Maggie: this would also invite extra feedback from the greater community.
+
+AI: link the zoom thing
+
+**Takeaways and action items**
+
+*   Rotation list for roles.
+*   Publish agenda items ahead of time, then add summary and action items for each of them, i.e. transcripts not necessarily needed.
+*   Link to Zoom so that anyone can  join.
+*   Publish short findings on specific topics, e.g. one pagers
+
+## Project board + open issues
+
+**Takeaways and action items**
+
+*   What’s the outcome of the group?
+*   Clarity around the role of AMP in the governance model (in particular wrt process flow).
+*   Clarify how we work with the TSC and other WGs.
+
+## Process update
+
+nainar@, AMP Project: from idea => launch in the open
+
+Link?
+
+**Takeaways and action items**
+
+*   Needs-[something]-review labels in github to  trigger work from others.
+*   Needs templates to lower the barrier of entry
+
+## CCPA
+
+**Takeaways and action items**
+
+*   AC to bring this to TSC (done right away) -> assigned to analytics WG
+
+## TSC meeting
+
+*   Implement stale bot to avoid losing issues
+*   Track issues in the AC’s GH tracker (linking to where the work actually happens)
+*   Ask WG to tag back to the AC’s GH issue
+*   => annual (at a minimum) AC blog post on what was accomplished this year
+
+## Feedback from ETSY
+
+**Takeaways and action items**
+
+*   Lots of improvements have been done, so re-running A/B tests would make sense.
+*   Raising custom analytics and  A/B frameworks to TSC
+*   Outreach, more documentation for custom analytics etc.
+
+Katie, staff engineer at Etsy
+
+AMP: did an experiment in 2017 with market pages, landing pages  designed to be  SEO friendly. Few issues: lack of JS at the time
+
+*   internal ads that sellers can bid internally on search terms, perf characteristics led to Ajax loads which we couldn’t do in  AMP.
+*   Internal analytics, homegrown A/B  testing
+
+Ran 3 buckets on SERP: AMP served from own server (not significant, no direct correlation found), from Google cache (20% decrease in visits), control.  Decided to stop.
+
+Q: did anyone look into  amp-script?
+
+A: don’t believe so. Main engineer that worked  on the AMP experiment left since.
+
+Hard  to justify maintaining  a fourth code base:
+
+*   Android app
+*   iOS app
+*   Web desktop and responsive
+
+Sumo: we ran a test too. 2017 analytics components was not reliable. Known issue. Would recommend to re-run the A/B test.
+
+Katie: is it possible to get data into our  backend?
+
+Maggie: there is a way. We have a customer who has a similar situation, e.g. iframe hack workaround. Amp-script might even make that better.
+
+Joe: this might indicate the need for better doc on these cases where the default components are not enough.
+
+Maggie: your internal ads could be implemented via amp-list, lots has changed.
+
+Elisa: seeing positive results.
+
+Maggie: was very common on initial experimentations, feature parity was the main explanations.
+
+Katie: time and manpower needed to generate and sustain different markups.
+
+Sumo: what do you mean?
+
+Katie: transforming regular HTML to AMP, restriction on tags, etc.
+
+Sumo: need to differentiate between landing page and higher interaction. Quality and depth of components  has dramatically improved.
+
+Sumo: we keep hearing pain points around custom analytics and A/B tests. We should raise this up to the TSC.
+
+Joe: what sort of KPIs do you look at?
+
+Katie: making it easy to design pages with built-in components, etc. Helping our engineers without having them to think about AMP or not AMP.
+
+## SXG, Web Packaging, ESCAPE
+
+Pablo: started early on. Problem: front end folks, but this is very heavy backend stuff.  Feedback is: when Akamai offers it, then we do it. Can’t do that on our end.
+
+Greg: heard similar feedback. Getting the certificate, signing, etc. Looking into providing the software to run on a cloud  instance, but would still require some plumbing.
+
+Pablo: folks  dealing with CDN are more like account managers, not actually implementing things.
+
+Sumo: wrote a one pager for IAB, domain verification is important for EC scenarios, payment on iOS, 50% of ecosystem. Similar issues on Android with paymentrequest api.
+
+3 questions:
+
+*   How hard is it to implement SXG
+*   Is SXG a good approach to solving the issue
+*   How do we get broader support for SXG at the browser level, iOS webkit in particular.
+
+[Explanation of the AMP viewer]
+
+Devlin: moving into Ark (a CMS), there is no configuration. Support for AMP in Ark is bad.
+
+David : +1 the notion that a CMS could offer support for SXG, policy decision of the site owner to have it.
+
+Matt: offer guidance  on how to do it; options
+
+David: SXG can  be turned key. Building AMP isn’t but that’s where differentiation happens.
+
+Ted: running services is a bit off-side for CDN, we couldhelp with managing certificates.
+
+David: image optimization, wasm; some of the benefits of SXG is that it limits who needs the key. With CDN, this could end up in multiple  pops.
+
+Ted: CDNs know how to manage keys.
+
+Greg: what could we provide to make it easier for CDN to  adopt  it
+
+Ted: optimization needs to happen before the signing.
+
+Tobie: does one need multiple package per cache?
+
+Gregg: at the moment yes but with subresource substitution it wouldn’t be needed.
+
+[Problem put to rest]
+
+David: which options  would be appealing to Vox?
+
+Joe: looking at all  the options, roll ourselves was the favorite, not enough value at the time. Going  through a  CDN would be better. 
+
+David: is it about the lack  of value or  the lot of work aspect?
+
+Joe: the latter. 
+
+Gregg: looking  into something that anyone could  deployed on a cloud service.
+
+Joe: this looks appealing.
+
+Gregg: challenges are certificate management and making it easy to deploy.
+
+David: 
+
+Joe: every page looking the same while swiping, is a bigger issue.
+
+Maggie: e-commerce is a clearer cases, continuity.
+
+Joe: paywalls, premium members
+
+Kenji: Yahoo Japan is excited about SXG for that reason: premium only  features, benefits.
+
+Ted: lots of CDNs are standardizing on edge compute. End of next year. JS.
+
+David: Fastly is going with  wasm.
+
+Ted: financial implication
+
+Gregg: security implications as those pieces of infra might not be designed, APIs to access would make sense.
+
+Ted: agreed.
+
+Kenji: SXG, web packaging and portals are also important for opening up to non AMP content, in places where only AMP content is accepted. Explaining the magic of AMP via web  platform APIs.
+
+David: request  for AC input
+
+*   what are the things that would help adoption, lower the bar in ways that aligns with the value? 
+*   What is the benefit of fixing the URL for publishers?
+
+[tobie]: https://github.com/tobie
+[wirelessjoe]: https://github.com/wirelessjoe
+[cvazac]: https://github.com/cvazac
+[mobtec]: https://github.com/mobtec
+[levidurfee]: https://github.com/levidurfee
+[sumodas]: https://github.com/sumodas
+[edent]: https://github.com/edent
+[senthilp]: https://github.com/senthilp
+[tones]: https://github.com/tones
+[kenjibaheux]: https://github.com/kenjibaheux
+[elibud]: https://github.com/elibud
+[pdelgadorodriguez]: https://github.com/pdelgadorodriguez
+[dymerrell]: https://github.com/dymerrell
+[jervay]: https://github.com/jervay
+[mjwettergreen]: https://github.com/mjwettergreen
+[melsumner]: https://github.com/melsumner
+[msteffan]: https://github.com/msteffan
+[TedShuter]: https://github.com/TedShuter
+[LJWatson]: https://github.com/LJWatson
+[aghassemi]: https://github.com/aghassemi