peter levine

peter levine.png

intro’d to Peter via this share by John on fb:

John: How could I not like this? We’re moving from cloud computing to the next big thing – edge computing

1 min – subtract something that’s important today and fill it with something else.. and you’ll start to think out of the box.. rather than sequential.. ie: we could do more cloud computing.. subtract something and fill it with other dimensions..

subtract all data that measures (money et al) .. fill it with self-talk as data..  as the day [aka: not part\ial.. for (blank)’s sake…]

not just.. out of the box.. and not sequential..a nother way.. to live for 7 bn.. we have no idea how edgy that would be

4 min – the change is that the edge is going to become a lot more sophisticated.. w/iot…  all objects created over next 10 yrs.. each a data center.. ie: on wheels.. with wings.. floating… . these devices collecting vast amounts of info and that info needs to be processed in real time.. no time to go back to central cloud to get processed like google search gets processed right now….. will obviate cloud computing as we know it..

6 min – besides devices getting more sophisticated.. *it’s really all about the data.. for first time in computing history.. we’re collecting real world data about our environment

*where we need to replace.. all current data.. with self-talk as data.. imagine that.. that would change everything.. because.. 7 bn people would wake up..

imagine that energy\ness..right now.. we have no idea… what we’re missing.. talk about edge..

7 min – up until now.. us typing in … or computers generating.. first time we’re collecting world’s data… where has to be real time where data is collected..

yes.. let’s go beyond cars.. et al.. and do that with people.. ie: hosting-life-bits

8 min – we are now returning to.. back to the future.. returning to edge intelligence.. distributed computing model..

9 min – increasing orders via users… if think about total addressable market.. i always thought it was # of humans.. iot drives to new dimension.. because computers no longer attached to humans..  so let’s say .. beyod 7 bn… trillions of devices.. that’s the opp/challenge..

what if it’s not.. what if it’s that instead we focus on the 7 bn.. self-talking

10 min – interesting: intersection of machine learning and all this end pt data.. i believe machine learning.. catalyzes the edge adoption.. massive amounts of info requires machine learning approach.. to acquire it.. machine learning .. running at the end point.. the cloud becomes a place where learning occurs.. last place of important info storage.. but much of processing will move out to the edge.. and that’s where most important decisions will be the edge..

good.. because humans are notoriously poor at making decisions

or so we think.. we have no idea..

perhaps we have the wrong take on what decision making means..

12 min – these elements way more helpful to us than we are on our own data

at edge: 1\sensoring  2\inference 3\ action

on loop being fast enough to win

good on loop.. bad on needing to win..

13 min – loop prioritized agility over power.. this new paradigm is agility over power..not more powerful but more agile..

that’s good.. i think that happens more with hosting life bits.. focus on diff data

a lot with how this… and how frameworks for fighter pilots have been developed..

wrong sequence Peter.. let’s get out of that box.. not boding us well..

14 min – cloud does have a person.. all about learning… occurs centrally and propogates back out to edge


1\ sensors.. ie: cameras..  in running shoes.. tell you to shorten stride…

we have to deal with humanity.. first.. not how well we run.. oy .. go deeper…

data is valuable.. but too much to send back to cloud.. so has to be done at edge

2\ inference – unstructured.. highly variable.. machine will extract relevance.. thru algos.. give us powerful task specific

3\ action – real time data decisions..

with you.. but shorten time between intention and action..  of 7 bn people.. now how one person ie: runs better

18 min – complex.. in cloud itself.. cloud becomes training center for all this info.. fundamental aspect for machine learning.. needs lots of data.. ie: autos collecting info..

go deeper

19 min – predictions: 1\ sensor data explosion kills the cloud.. data stuck at edge.. computing done at edge.. p2p computing model..   think of the security challenges.. connecting trillions of p2p devices.. will be one of the opp/challenges  2\ move to world of data centric programming.. ie: teaching everyone to code logic.. when deal w data.. don’t use logic.. if then.. coders will be more mathematicians.. so transfer of type of talent needed… new programming languages developed around notion of data anal …   3\ processing power increased and price goes down…

so .. imagine that machine learning.. isn’t for us to get better at ie: running.. transporting.. even mtn climbing.. i mean chris mccandless.. didn’t go into the wild because he was craving a better stride.. a better car.. faster data… he went because he couldn’t bear to stay

created a society johann

machine learning has to go beyond getting more/faster data.. imagine we are focusing on the wrong data..[too much ness…gets us to inspectors of inspectors.. and people creating tools to inspect the inspectors.. wrong data via machine learning.. could get you to forgoing the mountain .. for the steep decline on the treadmill.. because over the years.. of the machine learning algo/pattern ness.. for you to climb better.. will mostly likely cause you to lose your humane .. antifragility..]

antifragile defn

imagine that machine learning.. at its best for humanity/world.. for your children.. for you…. is to help us all become indigenous..



find/follow Peter:

link twitter

wikipedia small

Peter J. Levine is an american engineering executive and venture capitalist.

Levine earned a BS in engineering from Boston University in 1983, and worked at Spectrum Software. He then worked as an engineer on Project Athena at MIT, and attended the MIT Sloan School of Management in 1988 and 1989. From 1990 through 2001 he was an early employee of Veritas Software, beginning his career as a software engineer and ending as a vice president.

Levine was a general partner at Mayfield Fund from 2002 through 2005. Levine was also the CEO of Mendocino Software from incubation (at Mayfield) in 2003, along with other former-Veritas executives Steve Colman and Jeffrey J. Rothschild. After a third round of funding in 2007, and licensing its software to a few customers, Mendocino shut down quietly by March 2008.

Levine became president and CEO of Xensource in February, 2006. After working to grow the faltering company’s revenues, Xensource was acquired by Citrix in 2007 for $500 million, despite few revenues. Levine became a vice president of Citrix.

Levine taught marketing and sales at the Sloan School of Management in 2010 and 2011, and at the Stanford Graduate School of Business starting in 2012. In March, 2011, Levine became a partner at the Silicon Valley venture capital firm Andreessen Horowitz, leading the firm’s investments in enterprise software including data center technology, enterprise applications and mobile computing. He serves on boards of directors including Bromium from 2011, Actifio from 2011, Mixpanel from 2012, Udacity from 2012, and Onshape since 205. He became a member of the board of trustees of the National Outdoor Leadership School in 2013 for a term through 2019.