Papers We Love - Videos

Corrina Sivak on Dynamo: Amazon's Highly Available Key-Value Store [PWL Tokyo]

2026-05-07T02:37:37Z

Corrina Sivak presents the seminal Dynamo paper from Amazon in Tokyo on April 30, 2026.

Paper: https://www.allthingsdistributed.com/files/amazon-dynamo-sosp2007.pdf

This paper presents the design and implementation of Dynamo, a highly available key-value storage system that some of Amazon’s core services use to provide an “always-on” experience. To achieve this level of availability, Dynamo sacrifices consistency under certain failure scenarios. It makes extensive use of object versioning and application-assisted conflict resolution in a manner that provides a novel interface for developers to use.

Rylan Talerico on Zep: A Temporal Knowledge Graph Architecture for Agent Memory [PWL NYC]

2025-12-07T17:04:41Z

Watch on YouTube

We're please to present Rylan Talerico on Zep: A Temporal Knowledge Graph Architecture for Agent Memory - Paper: https://arxiv.org/pdf/2501.13956

----
Papers We Love NYC would like to thank Datadog for making the NY Chapter events possible. Learn more about a career at Datadog: https://careers.datadoghq.com/
---

Today's large language models (LLMs) are stateless: at test time, the scope of their accessible information extends only to their internally encoded knowledge and the tokens in their context window. AI memory has emerged to address this, enabling long-horizon continuity and user personalization in LLM applications by intelligently hydrating the context window before inference. Zep is a low-latency, temporally aware, graph-based AI memory architecture. Zep reports strong performance on the Deep Memory Retrieval (DMR) and LongMemEval (LME) benchmarks, and is among the most well-known architectures in the space today. Zep retrieves and reconstructs relevant information across histories exceeding 115,000 tokens, as demonstrated in LME.

Rylan Talerico is co-founder and CPO of Retriever, where he works on AI memory and personalization. A self-taught engineer who dropped out of high school to self-direct, Rylan founded Crate.fm, a cloud storage and collaboration platform for musicians, before starting Retriever. Outside of work, he loves reading, music, and running.

Michael Vaughn on EXE: Automatically Generating Inputs of Death [PWL NYC]

2025-10-28T00:58:12Z

Watch on YouTube

We're please to present Michael Vaughn on EXE: Automatically Generating Inputs of Death (read the paper: https://web.stanford.edu/~engler/exe-ccs-06.pdf)

----
Papers We Love NYC would like to thank Datadog for making the NY Chapter events possible. Learn more about a career at Datadog: https://careers.datadoghq.com/
---

Autonomous testing complements conventional testing by leveraging cheap compute to explore software state spaces and uncover “unknown unknowns” beyond human-written tests. It spans a spectrum from random-input fuzzing, which is fast but struggles with complex conditions, to symbolic execution, which uses SAT solvers to systematically reach hard-to-hit paths—though these solvers can become prohibitively slow on complex constraints. Exe strikes a balance through concolic execution: it runs bare-metal code on concrete inputs while instrumenting paths with logical constraints, invoking a solver only when needed to explore alternate branches. This approach combines the speed of concrete execution with the path-finding power of symbolic methods, avoiding the full cost of traditional symbolic engines.

Michael Vaughn (he/him) has a PhD in computer science from the University of Wisconsin-Madison, and is a senior software engineer at Antithesis, working on their hypervisor and fuzzer. He spent the better part of a decade doing research at the intersection of operating systems and programming languages, somehow managing to write concerning amounts of x86 assembly, C, Scheme, Haskell, and LaTeX, often in the same day. He has also worked as a pub trivia host, and loves board games, hiking, and reading.

Alex Weisberger on Perfomal [PWL NYC]

2025-03-18T23:28:21Z

Watch on YouTube

We're pleased to present Alex Weisberger on Performal: Formal Verification of Latency Properties for Distributed Systems

Paper: https://dl.acm.org/doi/pdf/10.1145/3591235

---
We'd like to thank Datadog for sponsoring the PWL NYC Meetup
Join the Pack! https://careers.datadoghq.com/
---

Performal is an approach for formally reasoning about latency in distributed systems. It provides a way to state and prove worst-case latency bounds in a way that abstracts over the complexity of real-world infrastructure. To deal with the messy details of physical reality, it provides a way to pair this with performance measurements so that end-to-end latency can be accurately estimated.

Alex Weisberger is a full-stack engineer currently working at Datadog on the Database Monitoring product. At work, he's been working on database health analysis and recommendations. His extracurricular tech interests are around applying formal methods to day-to-day work: things like property-based testing, model-based testing, TLA+, and simulation. His weekends are mostly filled with Nerf fights, Super Smash Bros., and various sports with his kids.

Alex Kesling on Apache Arrow DataFusion [PWL NYC]

2025-02-12T12:08:36Z

Watch on YouTube

We're pleased to present Alex Kesling on Apache Arrow DataFusion: a Fast, Embeddable, Modular Analytic Query Engine http://andrew.nerdnetworks.org/other/SIGMOD-2024-lamb.pdf

---
We'd like to thank Datadog for sponsoring the PWL NYC Meetup
Join the Pack! https://careers.datadoghq.com/
---

DataFusion represents a new category of database technologies: "deconstructed databases". It's a relational engine on which you can graft any language frontend (with SQL provided out of the box) and point it at any data you want.

How could one technology possibly do the same work as SQLite, PostgreSQL, Pandas, DuckDB, what-have-you all in one package? Let's read the paper and talk about it.

---

Alex Kesling is the cofounder of Empathic, a new company exploring the intersection of enterprise data infrastructure and agentic AI. He previously worked on Google's Search and Knowledge Graph infrastructure, VR/AR tools for teachers, and a plethora of consulting projects. Alex loves systems design, time series visualization, distributed systems, and that one turkey sandwich the day after Thanksgiving.

Yotam Bentov on Error Detecting and Error Correcting Codes [PWL NYC]

2024-12-13T11:29:21Z

Watch on YouTube

We'd like to thank Datadog for supporting the NYC Chapter of Papers We Love - https://www.datadoghq.com/blog/engineering/
---
The best way to contact the NYC chapter is via the #nyc channel in Discord: https://discord.gg/6gupsBg4qp
---

We're pleased to present Yotam Bentov on Error Detecting and Error Correcting Codes

Paper: https://zoo.cs.yale.edu/classes/cs323/doc/Hamming.pdf

Bell Labs in the late 40s and 50s was the cradle in which modern computing was conceived. From the invention of the transistor to the conception of information theory, the intellectual environment of Bell Labs was critical in creating our present-day technological world.

Richard W. Hamming was steeped in this atmosphere when he came up with a still-used technique to for detecting and correcting errors in transmitted data. The paper introduces a technique for identifying the location of an error inside a transmitted "code word." The paper then proves that this technique yields the "most efficient" possible coding scheme.

We will use this paper as both a departure point for understanding error correction and detection as well as the environment in which Hamming was working. We will close by ruminating on what the state of computing in 1950 can tell us about the future of computing more broadly.

Yotam Bentov is an engineering manager for Tulip's Platform team, where he works on distributed systems, databases, and product type systems. He gets particularly excited about the history of computing, home-cooking, and bicycles.

Nathan Taylor on Liquid Type Systems [PWL NYC]

2024-11-22T15:45:18Z

Watch on YouTube

We're pleased to present Nathan Taylor on Liquid Types. (Read the paper: https://goto.ucsd.edu/~rjhala/liquid/liquid_types.pdf)

---
We'd like to thank Datadog for making the PWL NYC meetup possible!
Join the Pack https://careers.datadoghq.com/
---

Type systems are great! When you write a well-typed program, you've an assurance that certain facts about your program will always be true; at no point is a string passed to a function that expects an tuple argument, nor will one that is supposed to return an array have a corner case where it actually returns an int. The more expressive your type system is, the more specific those certain facts can be - a type system that supports generics can specify what type the keys and values of an dictionary must be, but for example, still couldn't specify that each element is an even integer. We'd need a more expressive type system for that! Programming language researchers are interested in exploring different ways to add expressivity to type systems without sacrificing usability. Liquid Types, published at PLDI in 2008, combines traditional type theory techniques with SMT solvers from the automated theorem prover research community to form a fascinating point in the design space. We'll begin by discussing the broader context around the work, learn a bit about how to digest an academic type theory paper, and then work through the details of the paper with the help of a toy implementation of the technique.

Nathan Taylor is a PhD student at UT Austin, where he ponders the intersection of formal methods and distributed systems. Before joining UT, he spent a decade in industry hacking on systems software ranging from GPU simulators, to JIT compilers and garbage collectors, up to distributed databases and edge computing networks. He was additionally a developer at Microsoft Research, a lecturer at MacEwan University and the University of Toronto, and a repeat speaker at Papers We Love SF.

Yiduo Ke on Bin packing can be solved within 1 + ε in linear time [PWL NYC]

2024-08-30T12:27:44Z

Watch on YouTube

We're pleased to present Yiduo Ke on Bin packing can be solved within 1 + ε in linear time.

Read the paper: https://sci-hub.scrongyao.com/10.1007/bf02579456

The bin packing problem is a well-known optimization problem in theoretical computer science, in which items of different sizes must be packed into a finite number of bins or containers, each of a fixed given capacity, in a way that minimizes the number of bins used. The problem has many applications, such as filling up suitcases, loading trucks with weight capacity constraints, creating file backups, and chip design. Bin packing can be solved within 1 + ε in linear time by De la Vega and Lueker provided the first polynomial-time asymptotic approximation scheme for this problem.

Yiduo Ke is a 2024 summer research intern with Espresso AI working on scheduling algorithms to optimize Snowflake utilization. She is doing a PhD program in theoretical computer science at Northwestern University.

---

Thanks to Espresso AI (https://espresso.ai/) for making this meetup possible!

Eric Leung on Attention Is All You Need [PWL NYC]

2024-07-16T19:35:41Z

Watch on YouTube

We're pleased to present Eric Leung on Attention Is All You Need.

📜 Paper: https://arxiv.org/abs/1706.03762

Attention Is All You Need is a landmark 2017 paper considered by some to be a founding document for modern artificial intelligence, as transformers became the main architecture of large language models. At the time, the focus of the research was on improving Seq2seq techniques for machine translation, but even in their paper the authors saw the potential for other tasks like question answering and for what is now called multimodal Generative AI.

Eric Leung is a marketing data scientist at the Walt Disney Company working on marketing measurement. He enjoys learning by teaching, and contributing to open source projects like freeCodeCamp. In his free time, you can find him playing the board game Go, editing Wikipedia, and figure skating.

💬 Join us on the Papers We Love Discord - https://discord.gg/6gupsBg4qp

---

Thanks to Datadog (https://www.datadoghq.com/) for making this meetup possible.

Ori Bernstein on An Introduction to Bε-trees and Write-Optimization [PWL NYC]

2023-06-21T23:44:06Z

Watch on YouTube

We're pleased to present Ori Bernstein on An Introduction to Bε-trees and Write-Optimization. (Read the paper: https://github.com/papers-we-love/papers-we-love/blob/main/data_structures/b-trees-write-optimization.pdf)

Bε trees are a write optimized data structure, similar to a B-tree. They have a number of interesting properties and uses.

Ori was inspired by this paper (and others) to write a file system, and will explain what they are, how they work, and why this paper caught his attention and inspired him to create something.

Ori Bernstein is more than 97% human, or triple your money back.

---

Thanks to Datadog (https://www.datadoghq.com/) for making this meetup possible.

PWLTO#15 – Nate Smith on Dynamo: Amazon's Highly Available Key-Value Store

2023-05-23T20:22:18Z

Watch on YouTube

This talk was given at the 15th meeting of the Toronto chapter of Papers We Love, hosted at Shopify Toronto on March 2, 2016:

http://www.meetup.com/Papers-We-Love-Toronto/events/228977227/

Nate Smith presented "Dynamo: Amazon’s Highly Available Key-value Store" by Giuseppe DeCandia, Deniz Hastorun, Madan Jampani, Gunavardhan Kakulapati,
Avinash Lakshman, Alex Pilchin, Swaminathan Sivasubramanian, Peter Vosshall, and Werner Vogels:

http://www.allthingsdistributed.com/files/amazon-dynamo-sosp2007.pdf

A Case for Correctly Rounded Math Libraries

2022-10-21T12:00:50Z

Watch on YouTube

Papers We Love Conf 2022
https://pwlconf.org/2022/santosh-nagarakatte/
Transcript: https://pwlconf.org/2022/transcripts/Papers_We_Love_Santosh_Nagarakatte.txt

Santosh Nagarakatte / Rutgers University

This talk will provide an overview of the RLIBM project where we are building a collection of correctly rounded elementary functions for multiple representations and rounding modes. Historically, polynomial approximations for elementary functions have been designed by approximating the real value. In contrast, we make a case for approximating the correctly rounded result of an elementary function rather than the real value of an elementary function in the RLIBM project. Once we approximate the correctly rounded result, there is an interval of real values around the correctly rounded result such that producing a real value in this interval rounds to the correct result. This interval is the freedom that the polynomial approximation has for an input, which is larger than the ones with the mini-max approach. Using these intervals, we structure the problem of generating polynomial approximations that produce correctly rounded results for all inputs as a linear programming problem. The results from the RLIBM project makes a strong case for mandating correctly rounded results at least for any representation that has fewer than orequal to 32-bits.

----

Video Sponsor: Cerner - https://engineering.cerner.com/?promo=pwlconf2022

Sense and Structure: Towards a Textual Analysis of Software

2022-10-21T12:00:40Z

Watch on YouTube

Papers We Love Conf 2022
https://pwlconf.org/2022/zach-tellman/
Transcript: https://pwlconf.org/2022/transcripts/Papers_We_Love_Zach_Tellman.txt

Zach Tellman / Microsoft Research

Software is, among other things, a text. It differs, however, from most texts we're familiar with. It's constantly changing, and is collaboratively authored. It's only loosely ordered, if at all. Our reading of software is generally instrumental; we read it in order to write it. And, perhaps most importantly, software has explicit boundaries; our interfaces are perforations, allowing it to be safely torn apart and rearranged.

Unfortunately, the relationship between software and other texts has been, as yet, barely explored. Most analyses of software, even by those with backgrounds in literature or linguistics, treat it as an opaque compiled artifact. This talk, drawing on ideas from analytic philosophy, semiotics, and literary studies, will try to make the textual qualities of software more obvious, both to those who write it and those who analyze its effects.

----

Video Sponsor: Cerner - https://engineering.cerner.com/?promo=pwlconf2022
Caption Sponsor: Datadog - https://www.datadoghq.com/blog/engineering/?promo=pwlconf2022

Give the Drummer Some More: Advances in Breakbeat Analysis/Synthesis

2022-10-19T12:00:40Z

Watch on YouTube

Papers We Love Conf 2022
https://pwlconf.org/2022/jason-hockman/
Transcript: https://pwlconf.org/2022/transcripts/Papers_We_Love_Jason_Hockman.txt

Jason Hockman / Birmingham City University

In the late 1980s, popular electronic music emerged at the critical intersection between affordable computer technology and the consumer market, and has since grown to become the one of the most popular genres in the world. Ubiquitous within electronic music creation, digital sampling has facilitated the incorporation of professional-quality recorded performances into productions; one of the most frequently sampled types of recordings used in these genres are short percussion solos from funk and jazz performances—or breakbeats. While these samples add an essential energetic edge to productions, they are generally used without consent or recognition. Thus, there is an urgency for the ethical redistribution of cultural value to account for the influence of a previous generation of artists. This presentation will present an overview on the topic of breakbeats and their relation to modern music genres as well as current approaches for breakbeat analysis, synthesis and transformative effects developed in the SoMA Group at Birmingham City University.

----

Video Sponsor: Cerner - https://engineering.cerner.com/?promo=pwlconf2022
Caption Sponsor: Datadog - https://www.datadoghq.com/blog/engineering/?promo=pwlconf2022

Hashing it Out? Understanding Psychoactive Substance Use in Programming

2022-10-19T12:00:32Z

Watch on YouTube

Papers We Love Conf 2022
https://pwlconf.org/2022/madeline-endres/
Transcript: https://pwlconf.org/2022/transcripts/Papers_We_Love_Madeline_Endres.txt

Madeline Endres / University of Michigan

Psychoactive substances, which alter cognition and perception, have the potential to have positive and negative impacts on programming tasks. Such substances are widely used in software, both currently and historically, but this use is not always well understood. This talk covers the research on the intersection of programming and substance use, with a focus on my recent work on the prevalence, perception, and usage motivations for cannabis use in software. By exploring individual substance use motivations, social and cultural ramifications, and the impacts of organizational policy, this talk examines how substance use can permeate all levels of software development.

----

Video Sponsor: Cerner - https://engineering.cerner.com/?promo=pwlconf2022
Caption Sponsor: Datadog - https://www.datadoghq.com/blog/engineering/?promo=pwlconf2022

3GC Collaborative and Creative Content Generation in Game Design

2022-10-17T12:00:41Z

Watch on YouTube

Papers We Love Conf 2022
https://pwlconf.org/2022/m-charity/
Transcript: https://pwlconf.org/2022/transcripts/Papers_We_Love_M_Charity.txt

M. Charity / NYU Tandon

Procedural Content Generation (PCG) systems have become more and more prevalent in game design and development (especially in the indie game community), as seen in popular games such as No Man's Sky, Minecraft, Hades, Spelunky, The Binding of Issac, and many others. These systems create unique experiences and interactions for the player every time they play, ideally leading to a longer, deeper engagement with the game. Most PCG systems are based on one-way interaction, where the system generates content for the player based on internal rules set by the game developers. However, what if there was a way to have the player influence how the content is generated? What if the system could learn from the players, and adapt its design process in real-time? In this talk, I'll show how game developers can create that two-way feedback loop between the player and the back-end AI system to create content that feels more human-like, more challenging, and more engaging.

----

Video Sponsor: Cerner - https://engineering.cerner.com/?promo=pwlconf2022
Caption Sponsor: Datadog - https://www.datadoghq.com/blog/engineering/?promo=pwlconf2022

A Requiem for SIDH: Efficient algorithms for supersingular isogeny Diffie-Hellman

2022-10-17T12:00:38Z

Watch on YouTube

Papers We Love Conf 2022
https://pwlconf.org/2022/deirdre-connolly/
Transcript: https://pwlconf.org/2022/transcripts/Papers_We_Love_Deirdre_Connolly.txt

Deirdre Connolly / Zcash Foundation

Isogeny-based cryptography went from a curiosity to a serious post-quantum contender arguably because of this paper from Costello, Longa, and Naehrig from Microsoft Research. But excellent applied cryptography did not save SIDH from the design choice that made it useful, and doomed it. We will explore isogeny-based cryptography, how we got SIDH, how this paper was a game-changer for isogeny-based cryptography, how SIDH died, and what isogeny-based cryptosystems may pick up the baton for the future.

----

Video Sponsor: Cerner - https://engineering.cerner.com/?promo=pwlconf2022
Caption Sponsor: Datadog - https://www.datadoghq.com/blog/engineering/?promo=pwlconf2022

Jake Moshenko on Zanzibar: Google’s Consistent, Global Authorization System [PWL NYC]

2021-07-29T01:03:21Z

Watch on YouTube

Papers We Love NYC - June 28, 2021

Jake Moshenko, CEO of Authzed, presents Zanzibar: Google’s Consistent, Global Authorization System (https://research.google/pubs/pub48190/)

In this talk, Authzed’s CEO Jake Moshenko, will talk about what makes Zanzibar a paper worthy of dropping everything and forming a company around. Zanzibar is the singular global-scale authorization service that powers permissions and sharing across all Google properties, including Docs, YouTube, and Cloud IAM. The talk will cover how the paper lays out an engineer-friendly blueprint for building a highly scalable distributed system with flexible consistency guarantees. The talk will also cover details of some of the known existing implementations, and how and why they have deviated from the paper.

Jake Moshenko is an engineering leader that has been building infrastructure tools and services for over 15 years for large companies and startups alike. Having previously struggled to keep up with customer permissions demands on products such as CoreOS’s Quay and Tectonic, he was searching for a better way to do application permissions.

Michael Pigott on Toward a Generic Fault Tolerance Technique [PWL NYC]

2021-06-10T14:00:14Z

Watch on YouTube

We're happy to have Michael Pigott presenting Toward a Generic Fault Tolerance Technique for Partial Network Partitioning by Mohammed Alfatafta, Basil Alkhatib, Ahmed Alquraan, and Samer Al-Kiswany, University of Waterloo, Canada (https://www.usenix.org/conference/osdi20/presentation/alfatafta)

"When the network partitions, it partitions completely." While this is not a direct quote, nearly all distributed systems today assume that if a network partitions, the nodes in the network are separated completely, and nodes on one side of the partition cannot communicate with nodes on the other side of the partition. The authors behind "Toward a Generic Fault Tolerance Technique for Partial Network Partitioning" find this assumption to be inadequate, to catastrophic results. The paper presents a network communication layer which can detect partial network partitions, and re-route traffic so the entire system can remain available.

-----
Bio
-----

Michael Pigott is a student earning his master's degree in Computer Science at University of Illinois. I’ve been a professional software developer for over 15 years. I currently work on a microservices architecture in Go to offer online lending in the buy-now-pay-later space.

PWL Mini Panel with Adrian Colyer

2021-02-25T00:52:11Z

Watch on YouTube

This panel was part of a PWL Mini supporting USENIX:

https://paperswelove.org/2020/video/pwlconf-mini/

Adrian Colyer - Panel Moderator

Bio: Adrian Colyer is the author of the computer science blog The Morning Paper (), and a venture partner with Accel where he helps find and build great technology companies out of Europe and Israel. Prior to Accel he held technical roles at a number of companies including SpringSource, VMware, and Pivotal.

Site: https://blog.acolyer.org/

Ada Gavrilovska - Panelist

Ada Gavrilovska is an associate professor in the School of Computer Science at Georgia Tech. She directs the Kernel research group, focused on performance, scalability and efficiency problems across the systems software stack, including operating, distributed, and high-performance computing systems. Gavrilovska's research is supported by the National Science Foundation, the US Department of Energy, industry support from Cisco, Facebook, HPE, Intel, Intercontinental Exchange, LexisNexis, Samsung, VMware, and others, and the Applications Driving Architectures (ADA) Research Center, a JUMP Center co-sponsored by the Semiconductor Research Corporation and DARPA. She served as the program co-chair of the USENIX Annual Technical Conference in 2020.

Site: https://www.cc.gatech.edu/~ada/
Publications: https://dblp.org/pid/76/3229.html

Joe Hellerstein - Panelist

Joseph M. Hellerstein is the Jim Gray Professor of Computer Science at the University of California, Berkeley, whose work focuses on data-centric systems and the way they drive computing. He is an ACM Fellow, an Alfred P. Sloan Research Fellow and the recipient of three ACM-SIGMOD "Test of Time" awards for his research. Fortune Magazine has included him in their list of 50 smartest people in technology, and MIT's Technology Review magazine included his work on their TR10 list of the 10 technologies "most likely to change our world".

Hellerstein is the co-founder and Chief Strategy Officer of Trifacta, a software vendor providing intelligent interactive solutions to the messy problem of wrangling data. He has served on the technical advisory boards of a number of computing and Internet companies including Dell EMC, SurveyMonkey, Captricity, and Datometry, and previously served as the Director of Intel Research, Berkeley.

Site: https://dsf.berkeley.edu/jmh/index.html
Publications: https://dsf.berkeley.edu/jmh/publications.html

Dan R. K. Ports - Panelist

I am a researcher in the Systems Research Group at Microsoft Research.

My research focuses on distributed systems – using a combination of new algorithms and systems techniques to build practical systems that are faster, more reliable, easier to program, and more secure.

I take a broad view of the systems field: besides distributed systems, I've worked in operating systems, networking, databases, architecture, and security. I believe that looking across the entire systems stack yields interesting opportunities at the intersection of these areas.

Most of my work these days involves rethinking how distributed systems should be built for the datacenter environment. I lead the Prometheus project at MSR, which asks how we can use new reconfigurable devices, such as programmable dataplane switches and smart NICs, to support advanced systems applications. The key idea is to co-design distributed systems with new network primitives.

Before joining MSR, I was on the faculty in CSE at the University of Washington. I still advise a few excellent students over there. An increasingly long time ago (i.e., 2012), I was a student at MIT, where I was (approximately) Barbara Liskov's last Ph.D. graduate. Even before that, I was an undergraduate at MIT.

Site: https://drkp.net/
Publications: https://drkp.net/publications.html

Justine Sherry - Panelist

Justine Sherry is an assistant professor at Carnegie Mellon University. Her interests are in computer networking; her work includes middleboxes, networked systems, measurement, cloud computing, and congestion control. Dr. Sherry received her PhD (2016) and MS (2012) from UC Berkeley, and her BS and BA (2010) from the University of Washington. She is a recipient of the SIGCOMM doctoral dissertation award, the David J. Sakrison prize, paper awards at USENIX NSDI and ACM SIGCOMM, and an NSF Graduate Research Fellowship. Most importantly, she is always on the lookout for a great cappuccino.

Site: https://www.justinesherry.com/
Publications: https://www.justinesherry.com/papers.html

Hakim Weatherspoon - Panelist

I received my PhD in 2006 from the University of California, Berkeley, in the area of secure and fault-tolerant distributed wide-area storage systems (e.g. Antiquity, OceanStore, etc.). I received a B.S. in Computer Engineering from the University of Washington in 1999.

Site: https://www.cs.cornell.edu/~hweather/index.php
Publications: https://www.cs.cornell.edu/~hweather/publications.php

Irene Zhang on The Demikernel and the Future of Kernel-Bypass Systems

2021-02-25T00:51:10Z

Watch on YouTube

This talk was part of a PWL Mini supporting USENIX:

https://paperswelove.org/2020/video/pwlconf-mini/

The Demikernel and the Future of Kernel-Bypass Systems

Abstract: This talk presents the Demikernel, a new OS architecture for kernel-bypass I/O devices. Demikernel hides device complexity and heterogeneity by defining a new high-level, kernel-bypass I/O API and implementing it using different user-level, library OSes for each device type I will discuss the challenges in designing kernel-bypass library OSes and the future research directions in kernel-bypass for datacenter applications.

Paper: http://irenezhang.net//papers/demikernel-hotos19.pdf

Speaker Bio
--------------------
My research focuses on datacenter operating systems and distributed systems.

I completed my PhD in 2017 at the University of Washington, where I was advised by Hank Levy and Arvind Krishnamurthy. My thesis is on distributed systems for applications that span mobile devices and cloud servers. Before my PhD, I received my S.B. and M.Eng. from MIT and worked for 3 years in the virtual machine monitor group at VMware.

I was born in Beijing, China but spent most of my time growing up in Columbus, Indiana. My husband and I like to cook, travel and occasionally do computer science together.

Site: https://irenezhang.net/
Publications: https://irenezhang.net/publications.html

Toma Morris on CRaft: An Erasure-coding-supported Version of Raft [PWL NYC]

2020-09-03T11:39:22Z

Watch on YouTube

Paper: https://www.usenix.org/system/files/fast20-wang_zizhong.pdf

Authors: Zizhong Wang, Tongliang Li, Haixia Wang, Airan Shao, Yunren Bai, Shangming Cai, Zihan Xu, and Dongsheng Wang, Tsinghua University

This paper discusses a method for using erasure codes (specifically Reed-Solomon codes) to encode data stored in a Raft cluster to reduce the volume of storage and network traffic, while maintaining the failure-tolerance guarantees that Raft provides. We will briefly discuss the basics of Raft and erasure codes, and then talk through practical examples of how the cluster encodes, stores, and reconstructs data during different failure scenarios.

Speaker Bio
***

Toma Morris is a software engineering manager with the Trust and Safety team at Hinge where they and their team work to prevent scams, harassment, and violence. They have also worked in robotics, AI, medical devices, and various other kinds of projects. Toma studied both Electrical Engineering and History, and is as interested in talking about the early Islamic empire as in distributed systems.

David Ashby on Understanding Real-World Concurrency Bugs in Go [PWL NYC]

2020-07-22T22:50:33Z

Watch on YouTube

Understanding Real-World Concurrency Bugs in Go.

Authors: Tengfei Tu, Xiaoyu Liu, Linhai Song, and Yiying Zhang

This paper investigates whether go’s message-passing channel model leads to programmers making fewer concurrency mistakes by examining six real-world large-scale go code bases to see what sort of bugs they have to fix in practice. We’ll review the paper’s research and conclusions, dig into some of the code examples to make sure we understand how the various classes of bugs work, and take some side journeys into the history and influence of Hoare’s Communicating Sequential Processes, what happens when empirical research falls into the hands of its subjects, and the cutest little bunny mascot ever.

Paper: https://songlh.github.io/paper/go-study.pdf
Communicating Sequential Processes: https://www.cs.cmu.edu/~crary/819-f09/Hoare78.pdf

Speaker Bio
--------------------

David Ashby is a “Senior Systems Engineer” at SageSure Insurance, which means he works on distributed authentication and authorization services all day. Lucky him! He’s also an organizer of Papers We Love NYC and PWLConf and previously gave a lightning talk at PWLNYC on implementing toy versions of SHA1 and SHA256 in Ruby (https://www.youtube.com/watch?v=KWH7HmvmQ6o). Not bad for an English major who dropped out of his first college. When not fighting with computers, he posts pictures of his cat Cricket on Instagram.

Dan Bentley on Build Systems a la Carte [PWL NYC]

2020-05-30T17:30:00Z

Watch on YouTube

I love this paper because it explores a topic that rarely gets academic attention: build systems (such as Make). It takes a confused landscape and finds a meta-model that even the authors of those tools didn't comprehend. It makes a convincing argument that Excel (!!) is a build system. And it does it all with a healthy dose of functional programming/category theory.

I'll try to recap the paper, in terminology and figures that can make sense to users, not just functional programmers. I'll also try and lay out unexplored territory and questions I'm left with.

Authors: Andrey Mokhov, Neil Mitchell, Simon Peyton Jones

Paper: https://www.microsoft.com/en-us/research/uploads/prod/2018/03/build-systems.pdf

Dan Bentley is a Software Engineer who's currently CEO of Tilt. Tilt's a startup building a Distributed Developer Experience, aka Make for Microservices. Previously, he worked at Google on a build tool that was a predecessor to Bazel, Open Source, and Google Sheets. He's opened for The Who and has received two checks from Donald Knuth.

Papers We Love NYC would like to thank Datadog (https://www.datadoghq.com/) for making this meetup possible.

Catherine Holloway on The Mathematics of Origami [PWL NYC]

2020-05-30T16:00:24Z

Watch on YouTube

Origami designs can produce an amazing diversity of 3-dimensional shapes from a 2-dimensional piece of paper without cuts or glue. In the late 70s, David Huffman (of the Huffman Coding algorithm used in JPEG compression) showed how the same mathematical techniques used for quantifying the amount of charge in an arbitrary electric field, the Gaussian Sphere, can be used to prove which individual vertices in a crease pattern are possible from a simple piece of paper. Determining valid crease patterns has applications to computer graphics, materials science, and design.

Paper: http://www.organicorigami.com/thrackle/class/hon394/papers/HuffmanCurvatureAndCreases.pdf

Catherine Holloway is a Site Reliability Engineer in Finance. She previously gave a lightning talk about algorithms for lawn mowing, which has applications in computerized embroidery. She is about to become a mother and has already acquired the inevitable closet full of unused crafting supplies and equipment.

Papers We Love NYC would like to thank Datadog (https://www.datadoghq.com/) for making this meetup possible.

Sean T. Allen on Deny Capabilities for Safe, Fast Actors [PWL NYC]

2020-01-17T13:15:00Z

Watch on YouTube

"Deny Capabilities for Safe, Fast Actors (2015)" by Sylvan Clebsch et al. lays out the core novel idea behind the Pony programming language: reference capabilities. Reference capabilities take object capabilities, turn them on their head and then apply them to variable aliases.

Not sure what that means? Sean T. Allen explains Deny Capabilities and how Pony combines them with an actor based model -- to provide a programming environment that allows for data-race freedom and fearless concurrency.

- Paper: http://soft.vub.ac.be/AGERE15/papers/AGERE_2015_paper_10.pdf
- Pony: https://www.ponylang.io/

Sean T. Allen is a member of the core team for the Pony programming language and an all-around swell fella.

Papers We Love NYC would like to thank Datadog (https://www.datadoghq.com/) for making this meetup possible.

Life Beyond Distributed Transactions / Space-efficient Static Trees and Graphs [PWL NYC]

2020-01-03T00:44:05Z

Watch on YouTube

Jonathan Brown presents Life Beyond Distributed Transactions by Pat Helland.

Paper: http://www-db.cs.wisc.edu/cidr/cidr2007/papers/cidr07p15.pdf

This paper explores and names some of the practical approaches used in the implementations of large-scale mission-critical applications in a world which rejects distributed transactions.

Jonathan Brown is a software engineer at Wallaroo Labs (https://www.wallaroolabs.com/), a role that has allowed him to explore and learn about many interesting problems in the distributed systems space.

---

Vaibhav Sagar presents Space-efficient Static Trees and Graphs by Guy Jacobsen.

Paper: https://www.computer.org/csdl/proceedings-article/focs/1989/063533/12OmNx2QUHQ

In this era of bountiful disk space and unlimited cloud storage, it seems
ridiculous to claim that data takes up too much of it. But it does!

In this 1989 paper, Guy Jacobson (whose PhD thesis is credited with creating this field of research) introduces a set of techniques to create what he calls "succinct data structures", which approach the information-theoretical lower bound for space usage while still allowing useful operations to be performed on the resulting representation. They accomplish this by getting rid of pointers for linking data and operating directly at the bit level. These techniques are used to encode trees and planar graphs, and although the details have changed in the last 30 years, the fundamental concepts are very much the same and still the focus of active research.

Vaibhav (https://twitter.com/vbhvsgr) used to write web applications for a living. He still does, but he used to, too. When he’s not doing that he yells about functional programming and package managers on Twitter, re-racks bumper plates after using them, and trawls the internet for the finest space disco.

---

Papers We Love NYC would like to thank Datadog for making the November meetup possible. https://www.datadoghq.com/

Papers We Love SF - Johnathan Chiu and Bruce Spang

2019-11-22T04:02:57Z

Watch on YouTube

Mini:
Johnathan Chiu is an undergraduate at UC Berkeley studying Electrical Engineering and Computer Science (EECS). His interests are in robotics, data compression, and immersive technology. He is currently doing research on Neural Network performance at Berkeley.

Main:
Bruce Spang on "The Detection of Defective Members of Large Populations"

Bruce is a PhD student at Stanford, working with Nick McKeown and Mary Wootters. His research is a combination of internet networking and theoretical computer science, and has spent way too much time thinking about group testing. Previously he worked as a software engineer at Fastly.

Nyah Check on Serverless Computing: One step forward and two steps backward [PWL-SF] 10/2019

2019-11-21T02:28:25Z

Watch on YouTube

---
Serverless Computing: One step forward and two steps backward
http://cidrdb.org/cidr2019/papers/p119-hellerstein-cidr19.pdf

Abstract:

Serverless computing offers the potential to program the cloud in an autoscaling, pay-as-you go manner. In this paper we address critical gaps in first-generation serverless computing, which place its autoscaling potential at odds with dominant trends in modern computing: notably data-centric and distributed computing, but also open source and custom hardware. Put together, these gaps make current serverless offerings a bad fit for cloud innovation and particularly bad for data systems innovation. In addition to pinpointing some of the main shortfalls of current serverless architectures, we raise a set of challenges we believe must be met to unlock the radical potential that the cloud—with its exabytes of storage and millions of cores—should offer to innovative developers.

Nyah's Bio
---
Nyah Check
twitter: https://twitter.com/nyah_check
Personal site: https://nyah.dev

Nyah is a Software Engineer at Altitude Networks, a cloud security startup, working on the Backend and Infrastructure. He spends his day writing IAAC for AWS and GCP. He is a seasoned conference speaker haven spoken at OSCON and Gophercon. In his spare time he enjoys writing poetry and playing basketball.

Jana Iyengar on The death of an end-to-end internet (and a way forward) [PWL SF] 08/2019

2019-10-22T18:03:43Z

Watch on YouTube

---
Over the past two decades, the Internet’s runaway success has caused its end-to-end architecture to be eroded by an organic proliferation of interposers or middleboxes, such as NATs, firewalls, IDS devices, and performance enhancing proxies, which now pose serious roadblocks to the Internet’s evolution. Through studies and anecdotes, we will discuss the architectural (and political) implications of middleboxes in our beloved end-to-end architecture, and strategies for protocol development and deployment in this brave new world.

Papers (read or skim):
[1] Is it Still Possible to Extend TCP? https://conferences.sigcomm.org/imc/2011/docs/p181.pdf
[2] Fitting Square Pegs Through Round Pipes: Unordered Delivery Wire-Compatible with TCP and TLS. https://www.usenix.org/system/files/conference/nsdi12/nsdi12-final67.pdf
[3] The QUIC Transport Protocol: Design and Internet-Scale Deployment. https://ai.google/research/pubs/pub46403.pdf (edited)

Jana's Bio

Jana is a Distinguished Engineer at Fastly. He was previously a Software Engineer at Google, working on QUIC. He is an editor in the IETF’s QUIC working group. He is a recovering academic, having worked on Minion, LEDBAT, SCTP, and other transport projects in his past lives.

On the Expressive Power of Programming Languages by Shriram Krishnamurthi [PWLConf 2019]

2019-09-19T13:37:29Z

Watch on YouTube

PWLConf 2019 Link: https://pwlconf.org/2019/shriram-krishnamurthi/
Slides / Captions: https://github.com/papers-we-love/pwlconf-info/tree/master/2019/shriram-krishnamurthi

On the Expressive Power of Programming Languages
Shriram Krishnamurthi, Prof. of Computer Science, Brown University

Papers are like poems. Some are dazzling, some are pedestrian, some are insightful, and some reward long periods of quiet contemplation. They stir up an emotional reaction that goes beyond the strictly rational, and can often be deeply personal.

In graduate school, during a period of identity crisis, I came across Matthias Felleisen's “On the Expressive Power of Programming Languages”. At a time when the world was ruled by C++, I had immersed myself in Scheme, so I always looked skeptically at mainstream linguistic claims. However, the language wars seemed beyond rational discourse. So the idea that someone could take a concept as nebulous as “expressiveness&rdquo and formalize it was already a revelation. But the beauty of this paper goes well beyond that: it also lies in the cleanliness of the approach, the correspondence of the formalism to intuition, and the tautness of its execution.

It was the most stunning paper I had ever read, and remains so. It's like the poem that never leaves your soul.

Unfortunately, this paper may not be easy to read for the uninitiated: it depends on a certain amount of “cultural knowledge” of programming language theory. I hope to peel off some of those layers and help you, too, understand the paper — hopefully while preserving the joy and beauty I experienced.

References
------------------

On the expressive power of programming languages
Felleisen M.
(1991) Science of Computer Programming, 17 (1-3) , pp. 35-75.
https://www.sciencedirect.com/science/article/pii/016764239190036W

Biography
----------------

Shriram is the Vice President for Programming Languages at Brown University in Providence, RI, USA. He’s not, really, but that’s what it says on his business card. At heart, he's a person of ill-repute: a Schemer, Racketeer, and Pyreteer. He believes tropical fruit are superior to all other kinds. He is terrified of success, because he may be forced to buy a suit. He is known to interrogate his audiences to ensure they’re paying attention. So, be alert. You can read email later.

Twitter: https://twitter.com/ShriramKMurthi
Site: https://cs.brown.edu/~sk/
DBLP: https://dblp.uni-trier.de/pers/hd/k/Krishnamurthi:Shriram

-----------------------------------------------------------------------------------------------------------
Video Sponsor: Comcast (https://jobs.comcast.com/)
Captioning Sponsor: Two Sigma (https://www.twosigma.com/careers/)
-----------------------------------------------------------------------------------------------------------

Anonymity in the Bitcoin Peer-to-Peer Network by Giulia Fanti [PWLConf 2019]

2019-09-19T13:37:24Z

Watch on YouTube

PWLConf 2019 Link: https://pwlconf.org/2019/giulia-fanti/
Slides / Captions: https://github.com/papers-we-love/pwlconf-info/tree/master/2019/giulia-fanti

Anonymity in the Bitcoin Peer-to-Peer Network
Giulia Fanti, Assistant Professor of Electrical and Computer Engineering at Carnegie Mellon University

Recently, researchers have demonstrated deanonymization attacks that exploit weaknesses in the Bitcoin network's peer-to-peer (P2P) networking protocols. In particular, the P2P network currently forwards content in a structured way that allows observers to link users' Bitcoin addresses to their IP addresses. This is a substantial privacy vulnerability that extends to several other cryptocurrencies as well. In this talk, I will discuss how these attacks work, and how effective they are at deanonymizing users. I will also talk about countermeasures, including proposed modifications to the networking stack.

References
------------------

An analysis of anonymity in Bitcoin using P2P traffic by Philip Koshy, Diana Koshy, and Patrick McDaniel; FC 2014 (https://www.ifca.ai/fc14/papers/fc14_submission_71.pdf)

Deanonymisation of clients in Bitcoin P2P network by Alex Biryukov, Dmitry Khovratovich, and Ivan Pustogarov; CCS 2014 (https://arxiv.org/pdf/1405.7418)

Dandelion: Redesigning the bitcoin network for anonymity by Shaileshh Bojja Venkatakrishnan, Giulia Fanti, and Pramod Viswanath; Sigmetrics 2017 (https://arxiv.org/pdf/1701.04439)

Anonymity properties of the Bitcoin P2P network by Giulia Fanti and Pramod Viswanath; NeurIPS 2017 (https://arxiv.org/pdf/1703.08761.pdf)

Biography
----------------

Giulia Fanti is an Assistant Professor of Electrical and Computer Engineering at Carnegie Mellon University. Her research interests span the algorithmic foundations of blockchains, distributed systems, privacy-preserving technologies, and machine learning.

She is a fellow for the World Economic Forum's Global Future Council on Cybersecurity, and has received a best paper award at ACM Sigmetrics and an NSF Graduate Research Fellowship. She obtained her Ph.D. in EECS from U.C. Berkeley and her B.S. in ECE from Olin College of Engineering.

Site: https://www.andrew.cmu.edu/user/gfanti/
DBLP: https://dblp.org/pers/hd/f/Fanti:Giulia_C=

-----------------------------------------------------------------------------------------------------------
Video Sponsor: Comcast (https://jobs.comcast.com/)
Captioning Sponsor: Two Sigma (https://www.twosigma.com/careers/)
-----------------------------------------------------------------------------------------------------------

Is Program Analysis the Silver Bullet Against Software Bugs? by Karim Ali [PWLConf 2019]

2019-09-19T13:37:19Z

Watch on YouTube

PWLConf 2019 Link: https://pwlconf.org/2019/karim-ali/
Slides / Captions: https://github.com/papers-we-love/pwlconf-info/tree/master/2019/karim-ali

Is Program Analysis the Silver Bullet Against Software Bugs?
Karim Ali, Assistant Professor of Computing Science at the University of Alberta

Program analysis is the art of reasoning about the run-time behavior of a program without necessarily executing it. This information is useful for various real-life applications such as supporting software developers (e.g., bug-finding tools, code refactoring tools, and code recommenders) and compiler optimizations. Program analysis is also used to ensure complex software adheres to standards and regulations (e.g., medical devices, car industry, and aviation industry).

In this talk, I will discuss the three main properties that enable program analyses to be useful in practice: scalability, precision, and usability. I will relate that to various papers that have been published in the field of program analysis, as well as some of the work that my group has done. I will conclude with where I see program analysis research going and the challenges that we aim to solve in the field.

References
------------------

Strictly declarative specification of sophisticated points-to analyses by Martin Bravenboer and Yannis Smaragdakis (https://people.cs.umass.edu/~yannis/doop-oopsla09prelim.pdf)

Dimensions of Precision in Reference Analysis of Object-Oriented Programming Languages by Barbara G. Ryder (http://prolangs.cs.vt.edu/refs/docs/cc03.pdf)

Averroes: Whole-Program Analysis without the Whole Program by Karim Ali and Ondřej Lhoták (https://karimali.ca/resources/papers/averroes.pdf)

FlowDroid: Precise Context, Flow, Field, Object-Sensitive and Lifecycle-Aware Taint Analysis for Android Apps by Steven Arzt, Siegfried Rasthofer, Christian Fritz, Eric Bodden, Alexandre Bartel, Jacques Klein, Yves Le Traon, Damien Octeau, and Patrick McDaniel (https://orbilu.uni.lu/bitstream/1099)

Precise Interprocedural Dataflow Analysis with Applications to Constant Propagation by Shmuel Sagiv, Thomas W. Reps, and Susan Horwitz (https://link.springer.com/content/pdf/10.1007/3-540-59293-8_226.pdf)

Context-, Flow-, and Field-Sensitive Data-Flow Analysis using Synchronized Pushdown Systems by Johannes Späth, Karim Ali, and Eric Bodden (https://karimali.ca/resources/papers/spds.pdf)

Debugging Static Analysis by Lisa Nguyen Quang Do, Stefan Krüger, Patrick Hill, Karim Ali, and Eric Bodden (https://karimali.ca/resources/papers/visuflow.pdf)

Towards Better Inlining Decisions Using Inlining Trials by Jeffrey Dean and Craig Chamber (https://courses.cs.washington.edu/courses/cse501/06wi/reading/dean-lfp94.pdf)

Reluplex: An Efficient SMT Solver for Verifying Deep Neural Networks by Guy Katz, Clark W. Barrett, David L. Dill, Kyle Julian, and Mykel J. Kochenderfer (https://theory.stanford.edu/~barrett/pubs/KBD+17.pdf)

Analyzing Deep Neural Networks with Symbolic Propagation: Towards Higher Precision and Faster Verification by Pengfei Yang, Jiangchao Liu, Jianlin Li, Liqian Chen, and Xiaowei Huang (https://arxiv.org/abs/1902.09866)

Soot (https://github.com/Sable/soot)

IBM. T.J. Watson Libraries for Analysis WALA (http://wala.sourceforge.net/)

Biography
----------------

Karim Ali is an Assistant Professor in the Department of Computing Science at the University of Alberta. His research interests are in programming languages and software engineering, particularly in scalability, precision, and usability of program analysis tools. His work ranges from developing new theories for scalable and precise program analyses to applications of program analysis in security and Just-in-Time compilers.

Karim obtained his PhD degree from the University of Waterloo in 2014, and was a postdoctoral research at Technische Universität Darmstadt (TU Darmstadt) 2014–2016.

Twitter https://twitter.com/karimhamdanali
Site: https://karimali.ca/
DBLP: https://dblp.org/pers/hd/a/Ali_0001:Karim

-----------------------------------------------------------------------------------------------------------
Video Sponsor: Comcast (https://jobs.comcast.com/)
Captioning Sponsor: Two Sigma (https://www.twosigma.com/careers/)
-----------------------------------------------------------------------------------------------------------

What About the Natural Numbers? by José Manuel Calderón Trilla [PWLConf 2019]

2019-09-19T13:37:15Z

Watch on YouTube

PWLConf 2019 Link: https://pwlconf.org/2019/jose-trilla/
Slides / Captions: https://github.com/papers-we-love/pwlconf-info/tree/master/2019/jose-manuel-calderon-trilla

What About the Natural Numbers?
José Manuel Calderón Trilla, Research Scientist at Galois, Inc.

30 years ago Colin Runciman asked What About the Natural Numbers? Now, in 2019, we find ourselves in need of a successor to carry Runciman's banner. Despite major advances in type systems and the growing adoption of the slogan 'make illegal states unrepresentable', we often rely on Integers in cases where negative values have no meaning. Runciman's paper reminds us of a fact that we all know: the choice in types can change the nature of an API. Integers are often the default in many systems and APIs, often for no reason beyond programmer familiarity. In this talk we will argue two main points: that for many cases Natural numbers retain all of the positive aspects of the Integers with none of the negatives, and that when designing a system or an API we should constantly be asking ourselves "What about X?"

References
------------------

What About the Natural Numbers by Colin Runciman (https://www.sciencedirect.com/science/article/pii/0096055189900040?via%3Dihub)

Biography
----------------

José Manuel Calderón Trilla studied Music Engineering Technology at the University of Miami before pursuing graduate work in Natural Computation (MSc) and Programming Languages (PhD) at the University of York in the UK.

After his PhD work he became a Research Scientist at Galois, Inc. working mostly on Quantitative Information Flow, compilers, and abstract interpretation. His research interests are in parallel functional languages, DSLs, computer music (particularly the representation and manipulation of structures representing rhythm), and compilers.

Playing music and baking bread are what occupy most of his time, though he has yet to find a way to combine the two hobbies.

Twitter: https://twitter.com/josecalderon
Site: http://jmct.cc/
DBLP: https://dblp.org/pers/hd/t/Trilla:Jos=eacute=_Manuel_Calder=oacute=n

-----------------------------------------------------------------------------------------------------------
Video Sponsor: Comcast (https://jobs.comcast.com/)
Captioning Sponsor: Two Sigma (https://www.twosigma.com/careers/)
-----------------------------------------------------------------------------------------------------------

Building Personable Machines by Star Simpson [PWLConf 2019]

2019-09-19T13:37:11Z

Watch on YouTube

PWLConf 2019 Link: https://pwlconf.org/2019/star-simpson/
Slides / Captions: https://github.com/papers-we-love/pwlconf-info/tree/master/2019/star-simpson

Building Personable Machines
Star Simpson, Research Scientist at Galois, Inc.

References
------------------

Matching In-Car Voice with Driver State: Impact on Attitude and Driving Performance by Ing-Marie Jonsson, Clifford Nass, Helen Harris, and Leila Takayama, 2005. (http://www.leilatakayama.org/downloads/Takayama.DriverEmotion_2005_prepress.pdf)

Biography
----------------

Star Simpson is an engineer and an entrepreneur, who started out in the world of infusing personality one character at a time into robots & gadgets in the physical world as an undergrad researcher at the MIT Media Lab in Prof. Cynthia Breazeal’s Personal Robotics group. She also started an event series called PLIBMTTBHGATY, or “Programming Languages I Have Been Meaning To Try But Haven’t Gotten Around To Yet,” a light-hearted opportunity to engage with computers in a convivial setting.

Twitter: https://twitter.com/starsandrobots
Site: http://starsimpson.com

-----------------------------------------------------------------------------------------------------------
Video Sponsor: Comcast (https://jobs.comcast.com/)
Captioning Sponsor: Two Sigma (https://www.twosigma.com/careers/)
-----------------------------------------------------------------------------------------------------------

Distributed Consensus Revised by Heidi Howard [PWLConf 2019]

2019-09-19T13:37:07Z

Watch on YouTube

PWLConf 2019 Link: https://pwlconf.org/2019/heidi-howard/
Slides / Captions: https://github.com/papers-we-love/pwlconf-info/tree/master/2019/heidi-howard

Distributed Consensus Revised
Heidi Howard, research fellow at Cambridge University’s Department of Computer Science and Technology

Distributed consensus, the ability to reach agreement in the face of failures and asynchrony, is a fundamental and powerful primitive for constructing reliable distributed systems from unreliable components. For over two decades, the Paxos algorithm has been synonymous with distributed consensus. Paxos is widely deployed in production systems, yet it is poorly understood and it proves to be heavyweight, unscalable and unreliable in practice.

In this talk, we re-examine the foundations of how Paxos solves distributed consensus. Our hypothesis is that these limitations are not inherent to the problem of consensus but instead specific to the approach of Paxos. The surprising result of our analysis is a substantial weakening to the requirements of this widely studied algorithm. Building on this insight, we are able to prove an extensive generalisation over the Paxos algorithm.

References
------------------

Distributed consensus revised by Heidi Howard (https://www.cl.cam.ac.uk/techreports/UCAM-CL-TR-935.pdf)

The part-time parliament by Leslie Lamport (https://lamport.azurewebsites.net/pubs/lamport-paxos.pdf)

Paxos Made Moderately Complex by Robbert Van Renesse and Deniz Altinbuken (http://www.cs.cornell.edu/courses/cs7412/2011sp/paxos.pdf)

Flexible Paxos: Quorum intersection revisited by Heidi Howard, Dahlia Malkhi, Alexander Spiegelman (https://arxiv.org/abs/1608.06696)

A Generalised Solution to Distributed Consensus by Heidi Howard, Richard Mortier (https://arxiv.org/abs/1902.06776)

the morning paper: Distributed consensus revised - Part I (https://blog.acolyer.org/2019/05/07/distributed-consensus-revised-part-i/)

the morning paper: Distributed consensus revised - Part II (https://blog.acolyer.org/2019/05/08/distributed-consensus-revised-part-ii/)

the morning paper: Distributed consensus revised - Part III (https://blog.acolyer.org/2019/05/10/distributed-consensus-revised-part-iii/)

Biography
----------------

Dr. Heidi Howard is a research fellow based at Cambridge University’s Department of Computer Science and Technology. Heidi’s research focuses on improving consistency, reliability and performance in distributed systems. Heidi received her BA in Computer Science from Cambridge University in 2014.

In 2019, Heidi received her PhD from Cambridge University for her research on distributed consensus. Heidi is probably most widely known for her generalizations of the widely used Paxos algorithm for solving consensus, including her work on Flexible Paxos.

Twitter: https://twitter.com/heidiann360
Site: http://heidihoward.co.uk
DBLP: https://dblp.uni-trier.de/pers/hd/h/Howard:Heidi

-----------------------------------------------------------------------------------------------------------
Video Sponsor: Comcast (https://jobs.comcast.com/)
Captioning Sponsor: Two Sigma (https://www.twosigma.com/careers/)
-----------------------------------------------------------------------------------------------------------

Lightinging Round! 2019 [PWL NYC]

2019-09-19T11:44:16Z

Watch on YouTube

We've got a special edition of Papers We Love tonight as we round up a gang of speakers for Lightning Talks! Each speakers will present for 15 - 20m on a topic dear to their heart related to computers. We're still filling out the list of talks but you can see some of them below:

* Talk: Approximation Algorithms for Lawn Mowing and Milling - http://www.ams.sunysb.edu/~jsbm/papers/lawn-mow-cgta.pdf

Catherine Holloway almost completed a PhD in experimental quantum optics at the University of Waterloo's Institute for Quantum Computing but dropped out to work at a few robotics and quantum computing startups. She currently lives in Manhattan and is an SRE at Bloomberg.

* Talk: Branch Prediction and the Performance of Interpreters - Don't Trust Folklore (Erven Rohou, Bharath Narasimha Swamy, André Seznec, 2015). https://hal.inria.fr/hal-01100647/document

Max Marrone is a recent computer science graduate whose other interests include aviation and design. He aspires to someday be able to transfer laundry from the washer to the dryer without accidentally dropping wet clothing all over the dirty floor

* Talk: The Sensitivity Conjecture Solved!
http://www.mathcs.emory.edu/~hhuan30/papers/sensitivity_1.pdf

Homin Lee has a PhD in computational learning theory, and is currently a data scientist at Datadog.

* Talk: Flipping Bits in Memory Without Accessing Them: An Experimental Study of DRAM Disturbance Errors (Yoongu Kim, Ross Daly, Jeremie Kim, Chris Fallin, Ji Hye Lee, Donghyuk Lee, Chris Wilkerson, Konrad Lai, Onur Mutlu) - http://users.ece.cmu.edu/~yoonguk/papers/kim-isca14.pdf

Andrew Gross graduated with a degree in Applied Physics and the knowledge that he did not want to pursue a career in it. He currently lives in Queens and works as a Staff Engineer at YipitData.

------------------------------------------------------------------------------------------------------------
PWL NYC would like to thank DataDog for sponsoring this meetup!
https://datadoghq.com | https://www.datadoghq.com/careers
------------------------------------------------------------------------------------------------------------

Jessie Frazelle on A Tale of Two Papers [PWL SF] 08/2019

2019-09-03T05:57:46Z

Watch on YouTube

--
Mini
Kevin Burke on Baumol’s Cost Disease - On The Performing Arts: The Anatomy of their Economic Problems - http://people.stern.nyu.edu/wbaumol/OnThePerformingArtsTheAnatomyOfTheirEcoProbs.pdf

Kevin's Bio
Kevin Burke (https://twitter.com/derivativeburke) ( https://burke.services ) likes building great experiences. He helped scale Twilio and Shyp, and currently runs a software consultancy. Kevin once accidentally left Waiting for Godot at the intermission.

(Editor's note: we lost a minute or so of video at the start of Kevin's talk due to an issue with the AV system. Sorry Kevin!)

--
Main Talk
Jessie Frazelle on A Tale of Two Papers

"It was the best of times, it was the worst of times." Come dive into two papers covering datacenter outages: "Maelstrom: Mitigating Datacenter-level Disasters by Draining Interdependent Traffic Safely and Efficiently" and "What bugs cause production cloud incidents?" You'll laugh, you'll cry, you'll leave with the knowledge of knowing that we are all still just one bug away from disaster at any point in time.

* https://people.cs.uchicago.edu/~shanlu/paper/hotos19_azure.pdf
* https://www.usenix.org/conference/osdi18/presentation/veeraraghavan

Jessie's Bio
Jessie Frazelle, long time bug lover, gets a joy from debugging under pressure. Currently independent consultant. She couldn't decide on one paper so she picked two

Elizabeth Ramirez on Transition Matrix Estimation in High Dimensional Time Series [PWL NYC]

2019-07-27T15:07:29Z

Watch on YouTube

Elizabeth Ramirez (https://twitter.com/eramirem) will be presenting the paper "Transition Matrix Estimation in High Dimensional Time Series" (http://proceedings.mlr.press/v28/han13a.pdf).

About the Paper:

The state-transition matrix $A$ is a matrix you use to propagate the state vector over time, i.e. $x_{t+1} = Ax_{t} + Bu$. It usually comes from the equations that describe the system, but if you don’t know the dynamics of your system or the system is stochastic, this matrix has to be estimated. This paper presents multiple methods to estimate this matrix for high-dimensional VAR time series.

Speaker:

Elizabeth Ramirez is an Electrical Engineer and Applied Mathematician. I'm an Applied Scientist at Descartes Labs, where I solve very large linear systems that model complex systems, like transportation.

John Valois on Wait-Free Synchronization [PWL NYC]

2019-05-14T10:47:25Z

Watch on YouTube

How do we implement data structures in a shared memory environment? The conventional answer is to use mutual exclusion, but this approach does not behave well when we encounter delays or failures in the critical section, forcing other processes to wait.

Wait-Free Synchronization by Maurice Herlihy (https://cs.brown.edu/~mph/Herlihy91/p124-herlihy.pdf) explores an idea which ensures that operations complete in finite time regardless of the relative speeds of other processes. We’ll see a connection to the ubiquitous consensus problem and a framework for understanding what synchronization primitives are necessary and sufficient for implementing a given object, culminating in a method for implementing any object in a wait-free manner.

Bio:

John Valois is a Managing Director at BlackRock where he works on core platform engineering.

PWLTO#27 - Ben Darwin on State Machines All The Way Down

2019-05-01T12:07:33Z

Watch on YouTube

Ben Darwin presenting "State Machines All The Way Down" by Edwin Brady. Paper link: https://www.idris-lang.org/drafts/sms.pdf

This talk was given at the 27th meeting of the Toronto chapter of Papers We Love, hosted at StartWell on March 12th, 2019: https://www.meetup.com/Papers-We-Love-Toronto/events/259167965/

Sun-Li Beatteay on Guaranteeing Consensus in Distributed Systems with CRDTs [PWL NYC]

2019-04-18T11:37:14Z

Watch on YouTube

Consensus in distributed systems has been a debated topic every since programmers discovered they could run the same program on multiple machines. Researchers have been studying consensus for decades, resulting in numerous algorithms and white papers. Unfortunately, many of these algorithms are flawed and unreliable.

However, in 2011, a team of researchers published a paper on a novel approach to distributed consensus using Conflict-free Replicated Data Types (https://hal.inria.fr/inria-00609399v1/document). This paper created quite a buzz as it showed that CRDTs were mathematically proven to guarantee consensus through "Strong Eventual Consistency." They also claimed to have solved the CAP conundrum.

This presentation dives into this seminal paper in order to answer the hard questions. What are CRDTs? How do they work? And most importantly, does it actually solve CAP? By the end of this talk, everyone in the audience will have a foundational understanding of CRDTs and how they can be applied to their own work.

Best of all, I will be explaining all of this is as simple language as possible. No advanced math degree required! Sound too good to be true? You'll just have to come see for yourself!

Bio:

Sunny Beatteay (http://sunli.co) (@SunnyPaxos https://twitter.com/SunnyPaxos) is a software engineer and writer. He works on the Storage team at DigitalOcean where he helps to build Cloud products for fellow engineers. He also writes stories and tutorials related to technology on Medium (https://medium.com/@SunnyB).

Sunny lives in Brooklyn, NY. When he's not writing software, you can often find him drinking fine whiskeys, entertaining his cat, and playing Zelda. Usually at the same time.

Sarah Groff Palermo on Exception Handling: Issues and a Proposed Notation [PWL NYC]

2019-04-17T10:46:59Z

Watch on YouTube

Errors and debugging are the bane of a programmer’s life — and the source of many jokes, Twitter rants, and midnight breakdowns. As programming matures as a practice, we continue to add different ways to avoid and address errors, but how did we get here to begin with?

Exception Handling: Issues and a Proposed Notation from John B. Goodenough (1975) details the needs and goals of an exception handling system and then gets specific with suggestions of syntax, including remedies to known issues in the system.

In this talk, we will take a look at the development of one approach to errors, — throwing and handling exceptions — as it developed in PL/I. Some of these features have not made it all the way to common modern languages, so this is a chance to take a look at what we've lost to time.

Sarah Groff Hennigh-Palermo (sarahghp.com / @superSGP on Twitter) is an artist, programmer, and erstwhile data designer. Her work centers around methods to make computers and data more humane — more accessible, more flexible, more contextual. She has created data-obscured art sites, new computer languages, and hybrid nostalgia machines.

Her current focuses are livecode, digital abstraction, and discovering new ways to break things.

Sarah is an alumna of the School for Poetic Computation, Recurse Center, Brown University, and NYU Tandon School of Engineering.

John Feminella on Impossibility of Distributed Consensus with One Faulty Process

2019-03-05T06:06:45Z

Watch on YouTube

-----------------------------------------------------------------------------------
Sponsored and hosted by Two Sigma (@twosigma)
-----------------------------------------------------------------------------------

Description
------------------

If you think it's hard to get humans to agree on something, wait until you see how computers work! Computer scientists call this problem consensus, and when the computers involved are in an asynchronous environment, it's distributed consensus.

For about a decade prior to this paper, computer scientists had been debating whether distributed consensus was solvable in real environments. At the time, it was known that synchronous consensus, a weaker version of distributed consensus where everyone acts at the same time, was possible — and even better, it was resilient to crashed or unreliable processes. But was the same true for distributed consensus? Can you design a distributed consensus algorithm that is resilient to these kinds of failures?

As the paper's title ("Impossibility of Distributed Consensus with One Faulty Process") might suggest, the answer is "no, it isn't possible". This turns out to have sweeping implications for the reliability of distributed systems, as we'll see in our talk. See you there!

Bio
------

John Feminella (http://jxf.me/) (@jxxf (https://twitter.com/jxxf)) is an avid technologist, occasional public speaker, and curiosity advocate. He serves as an advisor to Pivotal (https://pivotal.io/), where he works on helping enterprises transform the way they write, operate, and deploy software. He's also the cofounder of a tiny analytics monitoring startup named UpHex (http://uphex.com/).

John lives in Charlottesville, VA and likes meta-jokes, milkshakes, and referring to himself in the third person in speaker bios.

André Freitas on Dapper, a Large-Scale Distributed Systems Tracing Infrastructure [PWL Porto]

2019-01-30T14:23:46Z

Watch on YouTube

Paper: Dapper, a Large-Scale Distributed Systems Tracing Infrastructure
https://ai.google/research/pubs/pub36356

Dapper is a distributed tracing system that Google developed to be able to diagnose all requests that microservices make to each other, originated by a single request (e.g. Searching something at Google). André will present the paper and the conclusions Google learned while developing this tool internally.

This paper inspired tools such as Zipkin https://zipkin.io/ and will help you understand what are the implications of using a Distributed Tracing System.

https://www.meetup.com/Papers-We-Love-Porto/events/cmmfnqyzcbtb

Time, Clocks and Ordering of Events in a Dist. System by Dan Rubenstein [PWL NYC]

2019-01-03T12:28:52Z

Watch on YouTube

Paper: Time, Clocks and the Ordering of Events in a Distributed System by Leslie Lamport - https://www.microsoft.com/en-us/research/publication/time-clocks-ordering-events-distributed-system/

Today, we take the ability to look at our phone and see what time it is for granted. But what if time weren’t so easy? In his 1978 paper “Time, Clocks, and the Ordering of Events in a Distributed System”, Leslie Lamport illustrates several key facts with time and events in distributed systems that have key importance for our ‘web-scale’ systems of 2018. We’ll investigate a few concepts from the paper, and try to understand why clocks may not be so straightforward after all.

Dan Rubenstein is a software engineer at Blue State Digital. He’s worked on political campaigns in three electoral cycles, and is happy to hear your theory for why the polls were wrong.

Divide and Conquer Algorithms by Elijah Ben Izzy [PWL NYC]

2019-01-03T12:28:16Z

Watch on YouTube

Paper: Divide and Conquer Algorithms for Closest Point problems in Multidimensional Space - http://www.cs.unc.edu/techreports/76-103.pdf

Given n points in k dimensional space, how can you efficiently find the pair that is closest together? It turns out that there’s an elegant, divide-and-conquer approach that utilizes a nifty trick. Jon Luis Bentley, a pioneer in the space of geometric algorithms, proposes this solution (and answers many more problems) in his original PhD thesis, written in 1976. The talk will focus in on his solution to the closest-pair problem, then discuss some general approaches to algorithm construction that he outlined when defending his thesis... all written with a type-writer.

Elijah Ben Izzy is a quantitative software engineer at Two Sigma. Ever since he started studying CS in college, he’s loved taking deep dives into complex, elegant algorithms and building out systems to support them. He came across Jon Luis Bentley’s PhD thesis when researching for an algorithms class, and found it to be a piece of archaeological computer science gold.

Problem Detection by John Allspaw [PWL NYC]

2018-12-09T15:57:16Z

Watch on YouTube

Published in 2005 in the journal Cognition, Technology and Work, "Problem Detection" explores the "process by which people first become concerned that events may be taking an unexpected and undesirable direction that potentially requires action." While this paper primarily centers on empirically rebutting previous theories of how problems are detected, it also puts forth many important observations and concepts for software engineering to pay close attention to. This talk won't just be a re-statement of the paper's core views; I will place these into a software engineering and operations context and connect them to SRE and DevOps worlds in ways that may be consequential.

The paper's authors are Gary Klein, Rebecca Pliske, Beth Crandall, and David Woods.

Paper: https://www.researchgate.net/publication/220579480_Problem_detection

Bio:

John Allspaw has worked in software systems engineering and operations for over twenty years in many different environments: biotech, government, online media, social networking, and e-commerce. John’s publications include the books The Art of Capacity Planning (2009) and Web Operations (2010) as well as the forward to “The DevOps Handbook”. His 2009 Velocity talk with Paul Hammond, “10+ Deploys Per Day: Dev and Ops Cooperation” helped start the DevOps movement.

John served as CTO at Etsy, holds an MSc in Human Factors and Systems Safety from Lund University, and is currently a Principal Researcher at Adaptive Capacity Labs.

A Tutorial on Thompson Sampling by Lydia Gu [PWL NYC]

2018-12-09T15:53:23Z

Watch on YouTube

Multi-armed bandits is an online machine learning framework which trades off exploitation, selecting the current best choice, and exploration, gathering data on unknown options. One strategy for implementing this tradeoff is Thompson sampling. First proposed in 1933 in the context of clinical trials, Thompson sampling was mostly forgotten in academic literature until the recent decade. Around 2010, a couple of papers demonstrated empirically its competitive performance, prompting a flurry of academic work. In this lightning talk, we will give an overview of the multi-armed bandits problem and the Thompson sampling algorithm, and see how it has been used by companies for personalization.

Paper: https://arxiv.org/pdf/1707.02038.pdf

Bio:

Lydia Gu is a tech lead at B12, a startup that's changing the way websites are made using humans + AI. She has an MEng from MIT and lives in New York, where she enjoys rock climbing, escape the rooms, and escaping the city.

Shape Decomposition for Multi-channel Distance Fields by Zach Tellman [PWL SF]

2018-11-14T03:14:21Z

Watch on YouTube

Papers We Love San Francisco 10/2018

---
Mini
Ramon Nogueira on BlinkDB: Queries with Bounded Errors and
Bounded Response Times on Very Large Data - https://sameeragarwal.github.io/blinkdb_eurosys13.pdf

Ramon's Bio
Ramon is a software engineer with a passion for making large systems easier to understand and operate. He currently works at Google on OpenCensus: an open source metrics and distributed tracing library for microservices. Previous hits include iCloud storage APIs at Apple, and startups in London and Johannesburg

Main Talk
Zach Tellman on Shape Decomposition for Multi-channel Distance Fields - https://dspace.cvut.cz/bitstream/handle/10467/62770/F8-DP-2015-Chlumsky-Viktor-thesis.pdf

Zach's Bio
Zach consults on the design of distributed systems and APIs. He has written "Elements of Clojure", a book which tries to put words to what most experienced engineers already know, and is working on a tool for exploratory data processing.