Marketing of Dask talks at events

[mikebeaumont]

There are a growing number of events (conferences, meetups, workshops, hackathons) where Dask, RAPIDS, and NVIDIA intersect. As the RAPIDS Product Marketing Manager, who works for NVIDIA, I want to start a discussion on how to market, promote, and collaborate on this work. I believe that the Dask project can benefit from broader exposure without compromising it's duty to the Dask community and open source.

As an example, @pentschev is speaking at Euro SciPy in September (https://pretalx.com/euroscipy-2019/talk/9DPFGM/). This talk discusses Dask, RAPIDS, CuPy, Numba, and GPUs. I think there's an opportunity for Dask to lead a promotion of Peter's talk on Twitter and tag the others for further amplification.

[shoyer]

👍 I would support promoting any talks that highlight dask through the project's social media channels.

[jrbourbeau]

I agree with @shoyer about using Dask's social media accounts (which, as far as I know, is just https://twitter.com/dask_dev) to promote talks or blog posts which highlight the project. It's already fairly common for the Dask twitter account to like or retweet other tweets which link to relevant talk materials or blog posts. Here are a few recent instances:

• https://twitter.com/reach_vb/status/1162988693453340672
• https://twitter.com/insideHPC/status/1158754194091700224
• https://twitter.com/stsievert/status/1149407244485844993
• https://twitter.com/HammanHydro/status/1134275348000956416

[mrocklin]

Thanks @mikebeaumont for bringing this up. And thanks @shoyer and @jrbourbeau for responding.

I think that @mikebeaumont 's post can be reduced to two questions:

1. Is it ok/a good idea to coordinate marketing efforts between affiliated companies like NVIDIA and the Dask marketing resources like the twitter account, blog, and so on.

   It seems like so far people are generally on board with this.

2. Operationally how does a marketing professional like @mikebeaumont do this coordination?

   It would be good to hear thoughts on how best to enable this.

One solution to the second part is for Mike to send me things that he wants us to tweet out (we work together) and I tweet them out. This puts me in a little bit of an odd situation personally and professinoally, because now I'm navigating between my employer and my role in the Dask community. It also leaves me open to seeming like a company tool, which would be bad both for myself personally, and for the project generally I think. I would love for us to find a way that didn't involve direct NVidia employees.

I'll also add a third question:

3. How do we extend this engagement/partnership not just to NVidia , but to other highly engaged companies as well, like Anaconda and QuanSight?

[mrocklin]

No one seems to be engaging here. @mikebeaumont is there anything else that you wanted to bring up here? Do any Dask maintainers have suggestions on the questions above?

[TomAugspurger]

My preference for handling these kinds of situations would be

1. The person promoting the content (say @mikebeaumont in this case) asks the maintainers of the twitter account to publicize something. This can be done in the dask-dev Gitter, or perhaps in a google-group / mailing list that all the owners of the dask-dev account are subscribed to.
2. A person without a conflict of interest decides to actually Tweet (or whatever) the content. And conflict of interest here can be pretty narrow I think (does the person work at the same company as the thing being promoted / requesting the promotion).

I think if we write up some clear guidelines about how the person Tweeting determines if something is appropriate then this should work well enough (for us. Would be curious to hear from @mikebeaumont if this would work from their point of view).

[jacobtomlinson]

Here's an over-engineered solution.

How about having a repo that behaves much like conda forge stages recipes repo. Someone PRs a text file containing the contents of a tweet. Content is iterated on using GitHub review features. On merge a bot tweets on our behalf and cleans up the repo.

This is a software engineers solution and appreciate that this may not appeal to everyone. I've had mixed success with using Jekyll in a multi-disciplinary team for the same reason.

[rgommers]

@jrbourbeau and myself discussed this with Troy, Quansight's marketing lead. We are supportive of the ideas in this issue about collaborating between Dask and companies engaged in its development, and would love to participate. @TomAugspurger's suggestion on how to handle this in practice seems simple and appropriate.

[mUtterberg]

Is the idea behind the repo-esque workflow to maintain transparency? If that’s not a concern, it might be easier to direct tweet requests to @dask_dev DMs.

[jacobtomlinson]

@mUtterberg yes that was my intention. Given that Dask maintainers work for different organizations it is important to get unbiased peer review.

However I also agree that this may not be a strong enough concern to warrant a technical solution. @TomAugspurger suggested Gitter which might also be a good place for discussion. Not all maintainers have access to @dask_dev DMs.

[mUtterberg]

Is there an avenue that is friendly to non-engineers submitting tweets? I’m not sure what the target group is, but something like survey monkey or google forms could be a way for people to submit tweets for cross-promotion in a non-PR way. I’m not exactly sure how that would fit with the repo/issue thought train, though.

[TomAugspurger]

I think most of us (Jim, Matt, myself) on the call were in favor of a low-tech solution. Someone with dask-related content contacts us, and we'll manually send it out.

As for how they contact us, our options are

• GitHub issue
• Gitter channel
• Email to a googlegroup
• Google Forms

[mrocklin]

Marissa, which of those options do you think would make non-software-developer folks comfortable? And which would you prefer personally?

On Thu, Sep 5, 2019 at 1:17 PM Tom Augspurger notifications@github.com wrote:

I think most of us (Jim, Matt, myself) on the call were in favor of a low-tech solution. Someone with dask-related content contacts us, and we'll manually send it out.

As for how they contact us, our options are

• GitHub issue
• Gitter channel
• Email to a googlegroup
• Google Forms

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/dask/community/issues/5?email_source=notifications&email_token=AACKZTH5Q7HBOE3OMREKZPLQIE5JVA5CNFSM4IN6RPLKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD6AAESI#issuecomment-528482889, or mute the thread https://github.com/notifications/unsubscribe-auth/AACKZTCEFYT7TJRANMLG6ZTQIE5JVANCNFSM4IN6RPLA .

[mUtterberg]

Even as a dev, I’m not familiar with gitter (we use Spark at work, so that’s where I spend most of my time).

IMO, the google form or email to a google group options sound equally accessible to non-devs & devs alike. Might be worth putting out a quick (Twitter?) poll to get wider feedback on that, though.

[mikebeaumont]

@TomAugspurger basic guidelines would be very helpful. Could we work through an example? How would you feel about promoting Adam's tweet/blog - https://twitter.com/adbreind/status/1170757153184829440?s=21. Can you walk through your evaluation process so I can understand any reservations?

[mikebeaumont]

I'm a fan of a simple google form that guides the community through this process. The more open, simple, accessible, the better in my opinion.

[TomAugspurger]

Thanks @mikebeaumont.

On https://twitter.com/adbreind/status/1170757153184829440?s=21 in particular

1. Adam apparently isn't affiliated with Nvidia, so I think any team member could choose to retweet it. That said, it's kinda cross-promoting rapids, so maybe it'd be better from someone outside Nvidia to retweet it
2. It's dask-related, so it's relevant
3. I skimmed through the content to make sure there isn't anything objectionable

So at that point I would be comfortable retweeting it. Thoughts on that process from other dask maintainers?

[mrocklin]

Copying the tweet here for convenience:

Go see @datametrician speak on @rapidsai tomorrow @ApacheCon if you can! Meantime, here’s my post on large-scale ML beyond Hadoop/Spark with @dask_dev and RAPIDS. Do more, easier and faster, with your valuable data!

This tweet is interesting because it has two separate messages. In this particular case, I'm in favor of the second line of the tweet, but not so hot on the first line. I personally probably wouldn't feel comfortable tweeting this because of the association with my employer, but I wouldn't be particularly against anyone else retweeting it.

[mrocklin]

I think that one of the reasons that the first line turns me off is that it instructs people to do something. I would be more comfortable if it said something like "I'm really excited to see ...".

I mention this just because it might be interesting to include in tone suggestions. Expressing facts, opinions, or states of excitement makes things feel safer to me than telling people to do things. In this case, obviously Adam doesn't actually expect people to follow his instructions, but the language is slightly closer to that end.

[pentschev]

I think that one of the reasons that the first line turns me off is that it instructs people to do something. I would be more comfortable if it said something like "I'm really excited to see ...".

I mention this just because it might be interesting to include in tone suggestions. Expressing facts, opinions, or states of excitement makes things feel safer to me than telling people to do things. In this case, obviously Adam doesn't actually expect people to follow his instructions, but the language is slightly closer to that end.

By no means to criticize the way you feel about these things, but I just want to say that this is very subjective, and I think no matter how you write it, some people will be more likely to enjoy one way or the other.

Only to offer an opposite point-of-view (mine): I don't ultimately care if he wrote the way he did, or the way you said you prefer ("I'm really excited to see..."), or a third way, as long as the message is clear, either way I don't think it would have changed the way I felt about the tweet as a whole. However, if I carefully stop to think about both (which I did because it's the subject here), I feel that the original tweet is more natural, in the way that it was just someone writing what came to mind, rather than carefully thinking each word. Anyway, I may not be the average marketing target too. :)

[mrocklin]

Sure. Fair enough.

On Wed, Sep 11, 2019 at 2:55 PM Peter Andreas Entschev < notifications@github.com> wrote:

I think that one of the reasons that the first line turns me off is that it instructs people to do something. I would be more comfortable if it said something like "I'm really excited to see ...".

I mention this just because it might be interesting to include in tone suggestions. Expressing facts, opinions, or states of excitement makes things feel safer to me than telling people to do things. In this case, obviously Adam doesn't actually expect people to follow his instructions, but the language is slightly closer to that end.

By no means to criticize the way you feel about these things, but I just want to say that this is very subjective, and I think no matter how you write it, some people will be more likely to enjoy one way or the other.

Only to offer an opposite point-of-view (mine): I don't ultimately care if he wrote the way he did, or the way you said you prefer ("I'm really excited to see..."), or a third way, as long as the message is clear, either way I don't think it would have changed the way I felt about the tweet as a whole. However, if I carefully stop to think about both (which I did because it's the subject here), I feel that the original tweet is more natural, in the way that it was just someone writing what came to mind, rather than carefully thinking each word. Anyway, I may not be the average marketing target too. :)

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/dask/community/issues/5?email_source=notifications&email_token=AACKZTCDJNBGS76NQRABVTLQJFSMVA5CNFSM4IN6RPLKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD6QASJQ#issuecomment-530581798, or mute the thread https://github.com/notifications/unsubscribe-auth/AACKZTGYTURFGY2LR3HYFE3QJFSMVANCNFSM4IN6RPLA .

[TomAugspurger]

As a trial: I've made a short google form at https://docs.google.com/forms/d/e/1FAIpQLSfnc20V6VFSvuU3SN4bym9oS-gbstUbMpg0zboN66ao42KnmQ/viewform?usp=sf_link that looks like

That'll be dumped into a google sheet, and the dask maintainers will be notified when the sheet is updated. Does that seem reasonable to people? Any other fields we should add (a "comments field for arbitrary notes from the submitter?")

FYI @mrocklin, the content in the example is https://blog.dask.org/2019/09/30/dask-hyperparam-opt. I wasn't able to see the tweets that you had scheduled.

[jacobtomlinson]

Might be nice to have a field for the submitter to suggest the full tweet too? They may want to highlight something specifically from the post or make a comment on it. It would also reduce the effort on the maintainer as they would not have to write the copy themselves.

There may also be multiple items of related content that could be highlighted together. Like this recent tweet for example.

Perhaps a freeform field for the whole tweet which you could just copy-pasta into twitter would be easier?

[datametrician]

As a trial: I've made a short google form that looks like

That'll be dumped into a google sheet, and the dask maintainers will be notified when the sheet is updated. Does that seem reasonable to people? Any other fields we should add (a "comments field for arbitrary notes from the submitter?")

FYI @mrocklin, the content in the example is https://blog.dask.org/2019/09/30/dask-hyperparam-opt. I wasn't able to see the tweets that you had scheduled.

I like this... I will test it out right now.

[datametrician]

I like this... I will test it out right now.

Or I won't... is there a link to this form somewhere?

[TomAugspurger]

Sorry @datametrician, I didn't publish the link.

https://docs.google.com/forms/d/e/1FAIpQLSfnc20V6VFSvuU3SN4bym9oS-gbstUbMpg0zboN66ao42KnmQ/viewform?usp=sf_link

[datametrician]

After trying out the google form a few times, it would be nice if there was an approved or denied response. I know I’ve been in limbo about a few, “did someone see it”, “was an action taken”, etc... where a simple thumbs up or down would be very useful.

[TomAugspurger]

I think I'm the single point of failure on the form right now. Added a few other maintainers.

I don't think there'd be a way to provide feedback to the submitter via Google Forms unfortunately.

On Wed, Nov 13, 2019 at 9:17 AM Joshua Patterson notifications@github.com wrote:

After trying out the google form a few times, it would be nice if there was an approved or denied response. I know I’ve been in limbo about a few, “did someone see it”, “was an action taken”, etc... where a simple thumbs up or down would be very useful.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/dask/community/issues/5?email_source=notifications&email_token=AAKAOISULFOIUZCLJNBKMWLQTQLA3A5CNFSM4IN6RPLKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOED6PILQ#issuecomment-553448494, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAKAOIVI4QWKE7QFK3KDCJTQTQLA3ANCNFSM4IN6RPLA .

[datametrician]

Adding more maintainers is great! Did you ever get a chance to review my last couple?

[jacobtomlinson]

Is this still the process for proposing tweets? Should I fill in the form or have we moved to something else?

[jacobtomlinson]

ping @mrocklin

[mrocklin]

I know that I personally am not tracking the form. Tom may be, but I don't think that this went anywhere.

I went to go add you in TweetDeck and it looks like you were already added, and that there is a pending request. Maybe try logging into tweet deck and clicking on Accounts in the lower left?

[jacobtomlinson]

Thanks @mrocklin. I'm more than happy to be able to tweet from the Dask account.

However in the case where I would want to tweet about RAPIDS or GPUs it would be appropriate to have a non-NVIDIA employee approve it.

[TomAugspurger]

I haven't been monitoring it recently. Can we just use this GitHub thread?

On Mon, Jul 20, 2020 at 3:52 AM Jacob Tomlinson notifications@github.com wrote:

Thanks @mrocklin https://github.com/mrocklin. I'm more than happy to be able to tweet from the Dask account.

However in the case where I would want to tweet about RAPIDS or GPUs it would be appropriate to have a non-NVIDIA employee approve it.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/dask/community/issues/5#issuecomment-660895790, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAKAOITYXGJAQWRL325AVOTR4QAVNANCNFSM4IN6RPLA .

[mrocklin]

+1 to using this github thread

On Mon, Jul 20, 2020 at 6:39 AM Tom Augspurger notifications@github.com wrote:

I haven't been monitoring it recently. Can we just use this GitHub thread?

On Mon, Jul 20, 2020 at 3:52 AM Jacob Tomlinson notifications@github.com wrote:

Thanks @mrocklin https://github.com/mrocklin. I'm more than happy to be able to tweet from the Dask account.

However in the case where I would want to tweet about RAPIDS or GPUs it would be appropriate to have a non-NVIDIA employee approve it.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/dask/community/issues/5#issuecomment-660895790, or unsubscribe < https://github.com/notifications/unsubscribe-auth/AAKAOITYXGJAQWRL325AVOTR4QAVNANCNFSM4IN6RPLA

.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/dask/community/issues/5#issuecomment-661045848, or unsubscribe https://github.com/notifications/unsubscribe-auth/AACKZTC2S4PCT4YKRE2YYE3R4RCKZANCNFSM4IN6RPLA .

[hugobowne]

We've got an event w/ @TomAugspurger, @mrocklin, and myself this Thursday on Scalable Machine Learning in Python.

Is this something that Dask could promote on twitter?

It wouldn't make sense for Matt to tweet it from dask_dev as it's a Coiled event.

https://twitter.com/hugobowne/status/1292824089531310088

https://www.eventbrite.com/e/scalable-machine-learning-in-python-science-thursday-tickets-116369928589

[martindurant]

Personally, I think it's OK to promote dask core devs in conversation about dask, even when hosted by a commercial company. The case of Coiled is a little special, so I can see why you would want to be careful - dask itself doesn't want to be seen as promoting or pushing people towards Coiled, but I think we can have wording clear enough.

[mrocklin]

Yeah, I think that it's ok too. What I think probably isn't ok is me making a unilateral decision to promote for-profit activity that benefits myself.

If people here want to say something like "The Coiled Science Thursday events seem to be community-based enough and in-line enough with Dask's marketing tone that we're comfortable with a blanket 'go ahead' policy in the future until rescinded" then I would go ahead and retweet by default, or add some community-based messaging around the events from the @dask_dev twitter handle. I think that I probably need blessing from other folks before doing that though.

[martindurant]

The Coiled Science Thursday events seem to be community-based enough and in-line enough with Dask's marketing tone that we're comfortable with a blanket 'go ahead' policy in the future until rescinded

I believe yes

[jacobtomlinson]

Seconded

[jsignell]

Yeah, I agree with Martin that conversations/work by reliable people about dask are worthy of a retweet.

[datametrician]

I support this, but my support may be seen as self-serving as RAPIDS uses Dask immensely and NVIDIA loves to promote RAPIDS.

[shoyer]

+1 for advertising this Coiled event on Twitter.

On Tue, Aug 11, 2020 at 3:13 PM Joshua Patterson notifications@github.com wrote:

I support this, but my support may be seen as self-serving as RAPIDS uses Dask immensely and NVIDIA loves to promote RAPIDS.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/dask/community/issues/5#issuecomment-672312478, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAJJFVU4XPMSZEIAR3GIHHTSAG7B3ANCNFSM4IN6RPLA .

[hugobowne]

We have another Science Thursday this week with @jacobtomlinson https://www.eventbrite.com/e/deploying-and-scaling-data-science-tools-on-distributed-systems-tickets-117357554605 From the above, it sounds as though you are all ok with dask_dev promoting these events. I'll ping Matt to do so. Feel free to comment with any concerns // questions.

[TomAugspurger]

I took care of it.

On Wed, Aug 19, 2020 at 12:21 AM Hugo Bowne-Anderson < notifications@github.com> wrote:

We have another Science Thursday this week with @jacobtomlinson https://github.com/jacobtomlinson

https://www.eventbrite.com/e/deploying-and-scaling-data-science-tools-on-distributed-systems-tickets-117357554605 From the above, it sounds as though you are all ok with dask_dev promoting these events. I'll ping Matt to do so. Feel free to comment with any concerns // questions.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/dask/community/issues/5#issuecomment-675857656, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAKAOIRENLKVOL7N5N3EWU3SBNONPANCNFSM4IN6RPLA .

[jacobtomlinson]

Thanks @TomAugspurger

[hugobowne]

I'm super excited about Science Thursday this week and it seems in line with the above so if @dask_dev could share, that would be awesome.

https://www.eventbrite.com/e/data-processing-at-blue-yonder-one-supply-chain-at-a-time-tickets-127808266967

Supply Chain Analytics! And "How to build a ML data pipeline at terabyte scale using parquet, dask and kartotek". Wow!

[hugobowne]

Happy New Year, all!

We have a Science Thursday with Eric Dill this week, a comparative analysis of Dask, Spark, and RAPIDS. If this seems in line with the above, a twitter share would be awesome: https://twitter.com/CoiledHQ/status/1344312292415066112

Thanks!

[jakirkham]

Happy New Year, all!

We have a Science Thursday with Eric Dill this week, a comparative analysis of Dask, Spark, and RAPIDS. If this seems in line with the above, a twitter share would be awesome: https://twitter.com/CoiledHQ/status/1344312292415066112

Thanks!

Happy New Year!

Sounds good to me. Though perhaps I'm biased. Maybe @TomAugspurger or @martindurant have thoughts? 🙂

[martindurant]

I have no objection

[jakirkham]

We are presenting on an image processing library to use with Dask ( https://twitter.com/quasiben/status/1380590542891986944 ). Curious if that would be ok to retweet 🙂

cc @quasiben