Pycon disappointment

[DavidA]

Do you mean the "official" presentations or the lightning talks? I
thought both were kind of bad. Jeff Rush was great in both of the
sessions I saw and the gaming presenters were also good. But I saw a
lot of people who had never presented and were unprepared. In fact,
one didn't have any code whatsoever to share and the other one only
started showing some code during the last 10 minutes of his time.

This was also my first time at PyCon and I thought I'd expand on what
Mike said as I feel pretty much the same way. I also want to provide
some constructive feedback that can hopefully help improve the next
PyCon.

I attended all the keynotes, 15 sessions and two days of the lightning
talks. I was disappointed with about one-third of the keynotes and
sessions. I found only a handful of the lightning talks interesting.
My biggest complaint was the lack of preparation of the speaker:

* in three cases the presenter had a recent problem with their laptop
but had no back-up plan (dead drive, dead power supply, unable to
get
video out to projector). The presenters didn't have a copy of their
presentation elsewhere (thumb drive, or even a printout) so they
just
winged it and the presentation was difficult to follow and
ineffective.
When I have presented at conferences in the past, we were required
to
submit our presentations and materials to the conference at least a
week before so they could make them available on a web site and also
on backup laptops at the conference.

* the PyCon feedback survey doesn't allow for any useful feedback
about
the presentations. You only get to pick your five favorites. There
should be forms available (hardcopy or online) where we can give
feedback
to the presenters themselves. My impression is that many of the
speakers
have presented at PyCon before and may do so in the future so this
feedback
can help them be more effective. I found it a bit ironic that I
attended
at least three sessions with a strong testing theme that talked
about
the importance of feedback in the development process and how it
helped
improve the quality of the final product, yet there was no channel
to
provide feedback to the presenters themselves. It seemed a glaring
omission to me that the PyCon survey had questions about whether I
shared
a room (who cares?) but not about the quality of the presenters and
presentations.

* As a PyCon first-timer, I was not aware of the open meetings and
BoF
discussions while I was there. I feel like I might have missed one
of the
more valuable parts of the conference simply because I was ignorant.
It
would have been nice to get the word out a bit more - maybe an
announcement
each morning at the beginning of the keynotes.

* There has been a lot of discussion about the reservation of
lightning talk
slots to sponsors. What I don't understand is why this wasn't
disclosed at
the conference. I've seen some of the organizers defend the
"experiment"
but no one explain why it wasn't mentioned beforehand. I'm left with
the
impression that the organizers knew this would be unpopular and
didn't want
to draw attention to it. I think a lot of this could have been
averted by
disclosing this change before the conference took place (in which
case the
community may have pushed back and convinced the organizers to
reconsider
the decision). Or at least it could have been disclosed at the
conference
so people could have decided to skip the lightning talks and
organize their
own ad-hoc meetings or talks. Experimenting isn't bad. But failing
to
disclose this information was a poor decision - especially at a
conference
that prides itself in openness and community involvement.

* Lastly, I found the technical depth at most talks to be too shallow.
I was
especially surprised at this because I've only been using Python for
two
years, so I still think I'm a bit of a noob. But if you looked
around at
the conference, you saw a bunch of people who are really into
programming
(so much that many of them were doing it _during_ the talks) so to
think that
the audience isn't capable of following deep technical discussions
is a bit
off the mark. At other conferences I've attended and/or presented
at, they
would typically rate presentations as a level 1, 2 or 3. I think
this would
help set people's expectations. That coupled with session-level
feedback, would
help the organizers plan future PyCon sessions that better match the
attendees'
interests.

That said, I did learn a few things at PyCon and found the overall
experience
pretty good. I simply had been hoping for a little more...

-Dave

 

[J. Clifford Dyer]

I just want to step in and offer my 2¢. This is my first PyCon, and I
agree that a lot of the Lightning talks seemed pretty useless. Overall
though, I had a great experience at this conference. I learned a lot; I
met a lot of cool people; and I got really excited about new ideas to
bring back home.

Django code lab was fantastic.

Teach me Twisted was a fantastic, innovative, and effective way to teach
a new technology. There was a little bit of difficulty hearing over the
cross-talk, but I just moved up front and had no further troubles (and
better access to the Balvenie single-malt!

Most of the sessions I attended were moderately to highly useful. FWIW,
none of my presenters had laptop troubles, except at teach me twisted,
but we weren't on as much of a time crunch, and we took care of that one
pretty easily and kept going.

The only useless one I attended was actually the most highly technical,
not because it didn't have good information, but because functions were
used without reference to what module they had been imported from and
slides containing 15-20 line functions were left up for about thirty
seconds, and then were gone. I couldn't even finish reading them.

Note to speakers: do not say

x, y = tee(foo)

say

from itertools import tee
x, y = tee(foo)

or better (for pedagogical purposes)

import itertools
x, y = itertools.tee(foo)

I don't disagree with the criticisms leveled throughout this thread, but
I do want to say that I think it has been a great conference, and for
me, the problems did not ruin the experience. Heed these criticisms and
it will be even better next year. Ignore them, and it will probably
degrade over time.

Thanks,
Cliff

 

[Dianne Marsh]

This was my first PyCon as well. I had heard glowing recommendations
about the lightning talks (from Bruce) previously, and I was really
looking forward to them. I, too, was disappointed.

I help to organize a community based conference, and we have struggled
with providing value for sponsors as well. I have some suggestions,
which I will offer here and to PyCon organizers. This sounds similar
to what one person described above, regarding how lightning talks were
managed in '07.

At CodeMash, we scheduled a daily slot for vendor sessions and clearly
marked them as such. We were concerned that attendees would simply
avoid the vendor sessions, which would backfire. To mitigate this
risk, we strongly encouraged our vendors to do something "different"
than a sales pitch for vendor sessions, asking them to consider
providing something meaningful for the audience. Talks weren't
reviewed; we just gave them a nudge when we discussed the vendor
sessions with them. They were entitled to choose a pure sales pitch
if they wanted to do so, but we definitely discouraged this activity.
And the sponsors responded with some great talks, and expressed
satisfaction in the entire process! The vendor sessions were well
attended, and it was completely transparent that they WERE vendor
sessions. I had been totally skeptical about providing vendor
sessions ahead of time, yet even *I* was won over. Vendors WANT
people to come to their sessions. Sometimes they, just like speakers,
simply need a little nudge in recognizing what makes a compelling
talk.

In my opinion, other speakers suffered from not knowing what makes a
compelling talk as well. I don't know what other talks were proposed,
but those that were on the schedule were often disappointing because
the speaker provided too much "background" and not enough "here's
what's cool" for me. Those were the talks that I walked out of. I
suffer from this same problem as a speaker and I'm trying to fix that
myself. I hope that other speakers are interested in doing the same.

As for the attitude that if you weren't involved with organizing
Pycon, you can't complain about it, that's a bit unfair. Several
people DID engage in the conference onsite, organizing Open Spaces
discussions (Bruce included). I saw Bruce both suggesting Open Spaces
talks and being recruited to convene them (and, in one case, even
reconvene one that had taken place earlier). That's being involved in
the process, and should not be discounted.

Furthermore, in my experience, people don't usually complain about
things that don't matter to them. It's important, IMO, to recognize
that the complaints you see on this group seem to come from the heart,
from a desire to see PyCon flourish and be a conference worth
attending. I certainly feel that way, and I suspect that the vast
majority of people who have offered constructive criticism here do as
well.

I'm bummed about the lightning talks at PyCon from 2008, but I have a
lot of confidence based on what I have read here from Jacob and
others, that things will be different in 2009. Thank you for
listening to the community feedback.

-- Dianne

 

[Jacob Kaplan-Moss]

I'm bummed about the lightning talks at PyCon from 2008, but I have a
lot of confidence based on what I have read here from Jacob and
others, that things will be different in 2009.

They will! This year's lightning talks[*] were disappointing because
nobody really thought through how having so many more sponsors changed
the dynamic. Now we know, and we'll fix it.

Jacob

[*] Personally, I thought the Sunday talks -- which featured no
sponsors -- were quite good. I think attendance was spotty because it
was the last day, and because Saturday's talks were so painful.

 

[Raymond Hettinger]

If the following seems unnecessarily harsh, it was even more harsh for
me to discover that the time and money I had spent to get to my
favorite conference had been sold to vendors, presenting me as a
captive audience they could pitch to.

My thoughts:

* Saturday and Sunday were much better than Friday.

* Open-source != Anti-vendor. The vendors are a vital part of the
open-source community.

* Lightning talks should be proportionate to their information
content. The most common vendor message can be done in 45 seconds
while the next speaker is setting-up: "Hi, I'm Raymond from Raymond
Enterprises. We sell Raymond's signature to Raymond's fans. We use
Python to crawl the web for clients wanting his signature. We're
hiring Python programmers with experience in web-crawling. We're
proud to sponsor for PyCon 2009. Good night and good luck.".

* The sole guiding principle for the conference should be whatever
best serves the attendees.

* As the conference gets bigger, lots of previously minor annoyances
will become more irritating. The conference organizers will adapt as
needed.

* Vendor/sponsor presentations should not have priority over
informational talks.

Also, lots of things went well:

* Sean stepped-in and fixed-up the wireless for Saturday and Sunday
(but on Friday the third-party wireless setup sucked mightily).

* The conference admin (checkin, schedule posting, etc) was excellent.

* The AV work was great (you'll soon be able to see HD recordings for
most talks).

* Steve Holden successfully created a new type of talk, "Teach me
Twisted".

* The feedback on the tutorials was excellent, the BoFs seemed to go
well, and the sprints are off to a nice start.

* The conference was close to the airport.

One last thought:

* Most of the conference work is done by volunteers. As the community
grows, more volunteers will be needed (for next year, I plan to help
by reviewing talk proposals).


Raymond

 

[dundeemt]

This is my third PyCon, and I've found a reasonably-sized cadre of

people who come for the hallway conversations plus a Bof or two,
having given up on hearing anything new, useful, or inspiring in the
talks. There are several people I know who would like to see a more
advanced academic track.

Yes, Yes. This was my third pycon and before, I had always left
feeling as though my brain had been stretched. (A good thing) This
years balance of talks and my choices didn't leave me with the same
feeling. I would have like to seen slightly longer talks, especially
the ones I liked

-Jeff Hinrichs

 

[dundeemt]

Well understood. Sorry if I implied it was an easy job. I know it
isn't.


This would be true, except that the two talks I proposed last year
were essentially denied because they were too advanced, so I didn't
even bother this year. Perhaps I should have, but the PERIOD needs to
at least be replaced by a COMMA as long as the talk-acceptance
committee continues to reject more advanced talk topics in favor of
HOWTOs and Introduction To Package X.

I agree - the balance wasn't as good. We can all agree that HowTos
and Intros are a necessary part of the conference talks track, but as
Robert pointed out some talks should be of a more advanced nature. I
enjoy those that stretch my brain. Alex M, Pyke and NetworkIO and
Mark Hammond's keynote were among my favorite talks.

-jeff hinrichs

 

[mahesh.prakriya]

If the following seems unnecessarily harsh, it was even more harsh for
me to discover that the time and money I had spent to get to my
favorite conference had been sold to vendors, presenting me as a
captive audience they could pitch to.

Yes, the keynotes were very boring compared to last year. If there's
only one thing ot change, I think sponsorship shouldn't entitle one to
a keynote.

FWIW, tho we sponsored at a Platinum level from Microsoft this year
but we declined to take up on any lightning talks, etc. To me, its
worth sponsoring PyCon (just for Python) irrespective of what we get.

 

[John DeRosa]

Note to speakers: do not say

x, y = tee(foo)

say

from itertools import tee
x, y = tee(foo)

or better (for pedagogical purposes)

import itertools
x, y = itertools.tee(foo)

I was scratching my head over tee() also, in the session where I heard
it. Were you in the "iterators II" session also? I've used itertools
a bit, but never tee(), and so when I thumbed through my copy of PER I
thought, ahh, I've skimmed over but never registered the importance of
that little bugger before... That was one of the more interesting
sessions to me.

John

 

[perrygreenfield]

Amen on the diamond keynotes and lightning talks. The lightning talks
were a great disappointment. Sponsor talks (or any such talks pitched
at selling or recruiting) should go in their own, clearly labeled
group so those of us who don't care about them can avoid them.

If there must diamond 'keynotes' put them at the end of a session or
in a separate track so we can easily avoid them if we wish. But
personally, I don't think keynotes should be for sale at all in any
form.

One problem I faced was that there were sessions that had few talks I
was interested in and other that had several at the same time where I
couldn't attend all that I was interested. It's likely that there is
no good solution to this, but perhaps one could try a new scheme for
scheduling talks by posting the talk list early and letting
registrants select the top n talks they want to see and running some
sort of scheduling optimizer that tries to satisfy most of these
desires (I have no idea if anything like this exists anywhere).

And if you do decide to change how you handle sponsorship don't be
afraid to say publicly how things are going to be different next time.
There could well be many who won't go next time (like me) unless they
have some reasons to believe that things will be different.

 

[Mike Orr]

But it gets worse. The lightning talks, traditionally the best, newest
and edgiest part of the conference, were also sold like commercial air
time.

We introduced sponsor lighting talks last year. This year it got out
of hand because there were twice as many sponsors. By the time the
Lightning Talk coordinators realized this, the sponsors had already
been promised a priority talk so we couldn't back out of it. So it
was a lack of foresight, not some commercial plot.

Next year we (the Lightning Talk coordinators) have recommended either
not having sponsor lighting talks, or moving them to a separate (non-
plenary) session. The vendor exhibition was much bigger this year,
and I think that's an adequate replacement for sponsor lighting
talks. If there are sufficient Open Space rooms, they can also create
their own session.

At first the morning plenary sessions -- where the entire conference
audience was in a single room -- just seemed a bit commercial. But
then I slowly figured out that the so-called "diamond keynotes" were
actually sold to vendors. It must have sounded great to some

I liked the mini-keynotes and I don't think they detracted from the
main keynotes. I did know what "diamond" meant so I knew they were
sponsor talks. I guess that should be clearer on the schedule.

What was supremely frustrating was discovering that the people wanting
to give REAL lightning talks had been pushed off the end of the list

The worst part of scheduling Lighting Talks is there's always more
interesting speakers than time. This seems to be an insolvable
problem.

The main problem I had at PyCon this year was the number of talk I
wanted to see that were scheduled at the same time as other talks I
wanted to see.

The highlight was the number of Open Space rooms and events. I didn't
attend any of these, but they seemed unusually lively this year.

On top of that, the quality of the presentations was unusually low.

I did feel that. An advanced track would be a good idea. Because you
do need to repeat stuff for the newbies. At least 30% of the
attendees were at PyCon for the first time.

--Mike

 

[fumanchu]

I agree - the balance wasn't as good. We can all agree that HowTos
and Intros are a necessary part of the conference talks track, but as
Robert pointed out some talks should be of a more advanced nature. I
enjoy those that stretch my brain. Alex M, Pyke and NetworkIO and
Mark Hammond's keynote were among my favorite talks.

Raymond Hettinger's talk on collections was not only one of my
favorites, it was apparently lots of other people's too--the room was
PACKED. I can't recall seeing any other talk that was even close to
seating capacity.


Robert Brewer
(e-mail address removed)

 

[George Sakkis]

vendors:

I did feel that. An advanced track would be a good idea. Because
you do need to repeat stuff for the newbies. At least 30% of the
attendees were at PyCon for the first time.

Not all first-comers are newbies; I attended for the first time too
but I've been using Python for the last four years or so. My overall
(totally unscientific) impression was that most attendants had at
least a decent grasp of the language.

George

 

[Mike Driscoll]

Raymond Hettinger's talk on collections was not only one of my
favorites, it was apparently lots of other people's too--the room was
PACKED. I can't recall seeing any other talk that was even close to
seating capacity.

Robert Brewer
(e-mail address removed)

The "Using PyGame and PySight to Create an Interactive Halloween
Activity (#9)" session with Mr. John Harrison was also quite full as
was the one for Pyglet. I think the nose presentation had people
sitting on the floor.

Geeks like games! I know I do!

Mike

 

[Jeff Schwab]

The "Using PyGame and PySight to Create an Interactive Halloween
Activity (#9)" session with Mr. John Harrison was also quite full as
was the one for Pyglet. I think the nose presentation had people
sitting on the floor.

Geeks like games! I know I do!

Me too.

As I have never attended PyCon, the amount of entertainment already
gleaned from this thread has wildly exceeded my expectations. Are
slides or notes from any of the presentations available online? What
was the topic of the well-received presentation from Google?

 

[NickC]

As I have never attended PyCon, the amount of entertainment already
gleaned from this thread has wildly exceeded my expectations. Are
slides or notes from any of the presentations available online? What
was the topic of the well-received presentation from Google?

I'm mostly intrigued by the tantalising hints being dropped regarding
Steve Holden's Teach Me Twisted talk

 

[Aaron Watters]

Amen on the diamond keynotes and lightning talks. The lightning talks
were a great disappointment. Sponsor talks (or any such talks pitched
at selling or recruiting) should go in their own, clearly labeled
group so those of us who don't care about them can avoid them...

Seconded. I haven't been at a Python Conf for a long time
but as a former attendee and (not very good) organizer I
have a couple suggestions based on my past experience and
mistakes:

- The conference is too long and it shouldn't be on the weekend.

- Almost all talks should be 10 minutes at most
with prepared slides and extended abstract with references.

- With much shorter talks you should be able to accept just about any
properly prepared talk (with abstract and slides) and this
should reduce the politics and increase the attendance (with
speakers and some colleagues and maybe broader interest).

I don't know about this conference, but in past conferences
I've been frustrated by people who give out a train of
conscience meander including popping in and out of various console
prompts, editors, web pages, guis... without conveying any useful
information (to me) in 30 minutes. If you tell them they have
10 minutes and make them get organized in advanced
they are much more likely to get to the point and
everyone can see something else before they run out of
attention span.

-- Aaron Watters

===
bye bye petroleum! good riddance.
http://biofuels.usu.edu/htm/initiative

http://www.xfeedme.com/nucular/pydistro.py/go?FREETEXT=pretty+boring

 

[atom.anderson]

OK, so why not get rid of all the talks and other stuff, and just have
a basically structureless conference, beyond scheduling some open
meetings on various topics? That would be a lot less expensive and a
lot more interesting.

For me as first time pycon attendee, i think this would be an absolute
disaster. The talks gave me an opportunity to sit next to new people
and meet people I wouldnt have otherwise if they had simply "put us
out to pasture" to chat it up.

I think for devs that are just meeting each other, having some sort of
subject matter to talk about is a big deal, and the talks forced
that. I agree though that once things get going, the hallway time and
BOF time would be fantastic.

-adam

 

[atom.anderson]

Okay!

I just read this entire thread to be caught up. I am a first time
PyCon-goer (as my previous post states). Because I have nothing to
compare this year's experience to, I'll give it to you as I saw it.
None of this is intended as a rant, (except maybe the lightning talk
section

Gripes
------

Numero Uno: The Lightning Talks.

The best lightning talk I saw was the one where the guy's code didn't
work and he couldn't believe it or simply move on with his
presentation, it was hilarious but I felt bad for the guy.

I have to be honest, I had heard GREAT things about the lightning
talks and I went to the session expecting to hear something great, or
at least feel the same sense of community I felt when discussing
python in education (i'm a student) or its use in the industry. I
went with a friend who also attended the conference from my school and
sat down expectantly.

I noticed the guy was trying to set up a powerproint (or OOo or
whatever) presentation and I simply couldnt believe it. A powerpoint
presentation? Pictures? text? preparation? That doesn't sound like
lightning at all. It sounds like, "slow-tning" talks. Methodically
prepared sales-pitch presentations on johnny q. coder's latest
commercial triumph. I thought these talks were spur of the moment,
community-delivered and off the cuff?

In my opinion, i don't believe that lightning talks should include the
option of using presentation software, maybe thats too restrictive,
but it seems to me that we'd be treated to much more grassroots or
user-given talks rather than sponsor-given ones.

I could just be ranting.


Number Two: Presenters should be required to post their slides on the
web site / schedule before they are permitted to present.

We want 'em, they've got 'em, and I was in more than one session where
simply having uploaded them to the PyCon schedule would have saved the
presenters bacon when it came time for their laptop to die or
something else.

I realize that these presentations are fluid and change (often the
night before!) but a failsafe like this wouldn't hurt anyone.

Number Three: Too much code, not enough concept.

Presenters this one's for you. I can't count the number of
presentations I attended where the presenter would click through three
slides of pycode just to show us a two or three-line snippet that
illustrated their point. Worse yet, it was often at the bottom of the
screen so no one but the front row could see it. This goes for text
two. I saw some great presentations as well, and they had limited
text on each slide. The last thing your audience wants to see is a
slide drenched in text of any kind.

You only have 30 minutes (or less). Show us brief couplets of code
for syntax sake. We don't care about code, we care about the
concept. If it were an hour lecture you could go through pages of
code and itd be great, but when we can't even read the whole slide
before you move on then its too much.

Number Four: What's a BOF?

Noob here. My first pycon and I didnt know the terminology. Shocker
huh? Heh, well i figured it out into the first day, but still didn't
quite get the concept. Around day two or three i started attending
these and learned just how cool they are.

With the RAPID growth pycon has shown over the last few years, perhaps
a little session at the beginning to help the newcomers or a
"Terminology" page in the handbook would be helpful?


Praise
------

As a student attending his first pycon, i must say it was AWESOME.
Depending on the sessions i chose to attend, i learned all KINDS of
useful stuff. Pyglet, Stackless (i'm willing to give it a chance
despite the mediocre presentation), RE, pysight and more.

Coming from a small school out west, the experience of visiting a
convention where a community actually existed was an incredible
experience. That was probably the best part of the conference was
seeing that there truly was a programming community and meeting other
like-minded people and being able to finally discuss things of
programming importance with them. I guess thats what we had hoped to
see more of in the lightning talks.

We enjoyed the EXPO as well and a couple of our graduating attendees
have nabbed phone interviews with companies that were represented
there.

I am definitely planning on returning to pycon next year, if only to
rub shoulders with other python programmers again and *hopefully*
attend a conference that has learned from any mistakes this year and
become an even better event for 2009.

adam

 

[Ed Leafe]

Number Three: Too much code, not enough concept.

Presenters this one's for you. I can't count the number of
presentations I attended where the presenter would click through three
slides of pycode just to show us a two or three-line snippet that
illustrated their point. Worse yet, it was often at the bottom of the
screen so no one but the front row could see it. This goes for text
two. I saw some great presentations as well, and they had limited
text on each slide. The last thing your audience wants to see is a
slide drenched in text of any kind.

This is good advice: simple slides serve as organization cues, but
the content should come from the speaker. The worst case (only saw
this twice at this year's PyCon) is when there is a text-heavy slide
that the presenter simply reads. We can all read it ourselves! Your
job is to elaborate on the topic.

I'd like to see two things regarding slides: first, if at all
possible, set a limit on the percentage of the talk that can consist
of slides. I would much rather see the presenter show actual
demonstrations of what they're talking about than simply talking about
it. If that's not possible, then in the session description, clearly
state the % of the talk that will be slides. Perhaps there are people
who like to sit in a room and watch long PowerPoint (-type)
presentations, but I'm not one of them. Let's see some code! Let's see
stuff working (and sometimes crashing!), and how changes affect the
results. When I've presented at PyCon and other conferences, that's
the part that I spend the most time on: preparing demonstrations. It's
not easy to do; certainly much more difficult than creating a slide
that sums up what the demo does. But it makes for a much more
interesting session!

-- Ed Leafe