HTM Hackers' Hangout (July 7, 2017)

live-event
community

#1

We skipped last month’s hangout, so I’ll be talking about a few things…

  • NuPIC 1.0 release update
  • switching NuPIC into maintenance for ongoing research
  • plans for sensorimotor integration visualizations and htm school
  • will try to drag in a research engineer to talk about ongoing research

You can watch here. I will post a link to join the call a few minutes before it starts.


#2

Is it at 9:00 AM out west, which would be noon here in the east?


#3

Yes, 12pm East coast.


#4

Join in for live chat right now here. To simply watch, go to to the YouTube page and we’ll start in about 20 minutes.


#5

Thanks for joining, @cogmission @Glenn_Gasner, @sheiser1, & @mrcslws.


#6

Wow, great to hear the update on the sensorimotor learning progress. I eagerly look forward to the paper and video. I will also check out Marcus’ htmresearch code to see what he’s come up with.


#7

13 posts were split to a new topic: How SMI might inform Temporal Pooling


How SMI might inform Temporal Pooling
How SMI might inform Temporal Pooling
#15

With the shift towards research, how is the new Numenta different from the former RNI.


How SMI might inform Temporal Pooling
#19

With the hackathons gone, will there be a return to the more research-y style workshops of 2008/2009. Or no community events at all.


#20

I’m still trying to figure this out. I’m willing to attend and speak at community events, but I’m not able to plan or execute them. But I would like to encourage people to have Meetups if you know there are multiple people in your area interested in HTM. We have had several community meetups in the past. I’d be happy to speak in SF or LA, but anyone could present in their town at a local meetup. If you do, please share any pictures or recordings!


How SMI might inform Temporal Pooling
#21

“Multiple people in your area interested in HTM” - LOL yes on my planet there are multiple people interested in HTM. With a subject that specialized you’d have to expect to travel, and I doubt anyone but Numenta would have the clout to organize something of this scale. The HTM workshops of the early days (BEFORE YOUR TIME MATT) were excellent.

I had high hopes for Numenta’s “open research” approach - mailing list, office hours, research github, commercialization etc. - to catch on and create a research community - a culture even! - along the lines of open source culture. That hasn’t happened really and even seems to be going backwards as of late.

And no just keeping this forum running isn’t enough of a thing. It’s 2017 and the internet still isn’t an adequate substitute for in-person communication.


#22

What else could help make this better? More office hours? More in-person events? If so, what kind? What else?


#23

For me this approach is working great. Like most good things, germination takes time. The best parts of HTM are still coming and the people interested enough to stick around will be there to witness their birth. Part of the beauty is that it really could be anyone who finds the next cool thing. This topic is more complicated than other hobbies, so it might have more turnover, but the rewards will be commensurate with the difficulty.

Personally, I believe that Numenta is the group who has it right. Other places will get there, but Numenta will get there first. The right minds are going to be drawn to the openness and accuracy here. There’s no way to participate in the theory or application at most other projects of this type without being hired and then hidden behind secrecy agreements.

There are always theoretical better alternatives to everything, but in reality this is the place that fits my processing best and I don’t even see a close alternative. Like any passion, this project isn’t going to be fed externally. You have to decide to love it yourself. The community here gives me a path for problem solving while learning. It also challenges me, personally, because I know the absolute most about the neocortex among my friends, and then I show up at Numenta discussions and I’m never the one with the insightful comment.

If someone is really waiting for the biggest group to form, show up right after the big discovery that’s coming eventually. There’ll be a day when everyone was suddenly “always” a fan. It’s okay to keep checking in and showing support with YouTube views until then.

Personally, I think the balance is being struck well between being open enough to feed the independent researchers, while being low burden enough that Numenta isn’t shutting it down because of the resource drain. I’m grateful for what exists and I’m trying to build a meaningful contribution. I’m grateful that when my project is done that there’s already an obvious place set up to share it, and an already-established community with enough background to understand why it’s great. It’s okay if the community ebbs and flows right now, as long as it stays alive through the lean times and stays focused through the rich times.


#24

Dunno. When Jeff said he wanted to be a ‘catalyst’ I assumed he had a plan. Maybe the world isn’t ready for this yet. In the end I’m not the one to ask, I’m a passive consumer only coming out of the woodwork because I’m sensing my cheese is about to be moved.


#25

Where do you live? I feel like there’s only a handful of us in Chicago even though I’m a novice.


#26

Glenn,

Good to hear you’re keeping your spirits up. How long have you been following Numenta? You see, this has been going on for ~10 years and the theory had great potential from day one. Zeta 1 alone could keep humanity busy for the next 100 years just applying it to pertinent problems. By now we should be having riots in the street with people demanding their share of the HTM pie and a UN conference on according human rights to AIs.

Instead what I saw at the last recorded office hour was 3 or 4 people hunched around a table somewhere in the corner of the universe, having a conversation mostly among themselves. Meanwhile my local pub around the corner has started having weekly meetups on AngularJS.

I’d lake to take it for granted that the Good Idea Always Wins and naturally this will live and succeed on its own merits. But then I remember reading “Guns, Germs and Steel” and how societies can devolve and the arrow of progress isn’t necessarily pointing forward all the time.

Well I guess should what’s left of the community dissolve and everyone goes their own way, it was nice meeting you all.


#27

Don’t make any assumptions. Our community is strong. There are a lot of people working on HTM projects. These ideas will stand the test of time.


#28

By now we should be having riots in the street with people demanding their share of the HTM pie and a UN conference on according human rights to AIs.

This imagery is great. It’s coming too, it just was always going to take more than ten years. Once we have really strong machine intelligence things’ll move faster, hopefully.

It’s okay to be frustrated with the progress here. Science has been a disappointment in my life in the raw number of huge breakthroughs. We have 7 billion humans, super fast computers and worldwide communication, and still cancer and heart disease have teeny-tiny progress since I was born decades ago. There are hopeful things on the horizon for those, but they’ve been 10 years out for the last forty years too.

There were solar panels on the Whitehouse in the 70s because that technology was going to take over electricity during the Carter administration. The 10 year predictions then still haven’t quite happened 40 years later, but they’re finally close. The great thing about 10 year expectations is that they make today happy because of the excitement of the thing that’s “about to arrive.” The only negative is that somewhere around 9.5 years people realize that they’re still about 10 years from that goal still.

The way to be happy here now is to “know” that this technology is getting incredibly close to huge discoveries which are going to be very fun to participate in. If you’re burnt out, the thing to do is to find something that has a ten year prediction you can believe in to make yourself happy today. The fact that the last ten years didn’t bring conscious AI doesn’t mean that conscious AI won’t be easy ten years from today. If you can feel the happiness of that thought and use it to be happy now and motivated now, then believe. If you can’t, then take a Numenta break and focus on whatever part of life fills you with excitement again. It’s even okay to switch to whatever is most popular all the time and always be a part of the biggest group supporting the newest thing. Find what brings you joy and do that. There will always be a core here that finds joy in this project.

I think something huge and fun is coming for HTM in 2018. I’m totally pumped to be here before it happened, so I get to be a witness to the moments beforehand when everyone had lost all hope. I think your post about the hopeless here will be part of the movie made about this time, to show that it wasn’t obvious before it happened that the big, huge, fun thing was so, so close.


#29

I plan on having more in-person events once the current research cycle is over and we have some more actual code APIs to show off. I imagine we might have some more hackathons or workshops at that point.