#implementing and algorithm would suck yeah but it's like THIS is the thing that makes you mad

[cliveguy]

really interesting venn diagram today between people acting like ao3 being down is the end of the world and people going full bloggers grab your laptops over tumblr implementing some new bullshit change lol

[yukipri]

I joined Threads! Unfortunately my reach on Insta is still the most limited of all of my socials, and that’s the account Threads is directly attached to.

If you’re on either Insta or Threads, would love if you could stop by🙏

Direct linking to Threads doesn't seem to work...so here's

my Instagram

through which you can find my Threads account!

Edit: You actually can link to Threads, ironically through the icon at the top of individual Instagram account pages on desktop. Though Desktop viewing is currently limited to individual pages, and you can't necessarily scroll the feed. So here's my Threads page!

~~

Next, some initial thoughts on the platform for those who are kinda curious what this new thing is. Not at all an expert, just some info I've gathered myself, that I'd personally have liked to know before joining:

On Threads

・It launched July 5~6, 2023. Rollout varies by country.

・Twitter's already upset about it, apparently there's a lawsuit. So I guess it considers Threads a possible threat.

・It's run by Meta, same folks as Insta and Fb.

・The accounts are directly linked to Instagram, which has its pros and cons. You need an Insta account to make a Threads account, but if you already have an Insta account, it's very easy to make a connected Threads account. However, once linked, they're linked; you can't disable your Threads account without also disabling your Insta account. You can also automatically follow all the accounts you're already following on Insta, to be activated as they make Threads accounts. You can also share Threads posts to your Insta stories.

・Currently only on mobile app, hoping the'll make it desktop accessible soon

・The current audience mostly seems to be folks who have come direct from Instagram. This means yes, a lot of official/influencer accounts, but also a lot of artists/small creators too.

Functionality

・Right now, the "feed/dashboard" is a hot mess. It's just one massive algorithmic spew, no way to curate it, no way to see just the posts by the folks you're following, not even chronological. These are things that i personally require in a social media, so yeah it sucks rn. HOWEVER, they did state that they're working on a "followers only" feed, much like the one on Instagram. If they implement a Followers Only feed like Insta, it'll at least be much better than Twitter's current "Following" feed. There has been no timeline provided on when this may launch.

・It functions much the same as Twitter, with options to create new posts, and like, retweet, quote retweet, and reply to posts made by others. (There's still a lot of floundering about the exact terminology to use, since it's not twitter so you can't "retweet" etc)

・500 Character word limit per post, which is much more than Twitter, ey!

・Up to 10 images per post, doesn't seem to cause image quality reduction either. Rather than being able to arrange the images like on Tumblr though, they're all automatically in a carousel that you can see by swiping.

・Your bio can be the exact same as what's on your insta, but currently only shows followers. In order to see who you follow, you tap "followers" and then when the menu opens, swipe to "following." Then, you can tap on the individual accounts to view their pages. This is currently the only way to ensure you see the posts from certain accounts, because there's no guarantee they'll show up on your feed, and it's a huge pain. But again, they're hopefully working on a fix for this.

My understanding is that the app rushed to launch now, since Twitter is a dumpster fire and people are fleeing. (why is it a dumpster fire? Well, Musk implemented a randomly fluctuating tweet view limitation, which is dumb af, but means it's essentially become unusable to folks who use the platform a lot)

I'm hoping that there'll be improvements to Threads and it'll become a viable Twitter alternative, but until then...yeah we'll keep an eye on it.

Now that I've probably completely turned you off from downloading the app, if you do, a nudge that again I'd appreciate if you stopped by my account, YukiPri_Art on Instagram, and/or YukiPri_Art on Threads!

[notaplaceofhonour]

see but I actually think this illustrates exactly the problem with a lot of GenAI discourse. this is less a case of unlikely partnership from disparate perspectives & more just people rebuilding religious (especially Christian) belief systems in secular/progressive language.

I have issues with a lot of the current implementations of GenAI—most of which stem from their origin in startups with a “move fast & break things” mentality leading to unsustainable & unethical practices—and (as a relatively new technology) the output often isn’t even good. but so much of the discourse is centered around an insistence that GenAI isn’t “real art” because xyz, and the arguments tend to boil down to “true art can only come from the divine spark of human essence” just rephrased in a vaguely secular way.

and (as a “Real Artist™️”) I just… fundamentally disagree with this need people seem to have to define art in an anthropocentric & pseudo-mystical way, centered around some undefinable human essence, or “originality”, or even intentionality.

obviously, GenAI is not actual “artificial intelligence” in the sense of a literal thinking machine—it’s essentially (and I know I’m oversimplifying) just a complex algorithm of machine learning/procedural generation that creates hallucinatory images—“A.I.” is just branding. but I have zero philosophical problem with the idea of some future artificially-created consciousness (or even an animal) creating things it would consider art, regardless of whether they have a soul/spirit, or whether their creation appeals to human sensibilities about what art requires or “should” be.

and tbh, independently of A.I. discourse, I think a lot of artists overestimate originality & the role of intentionality in art. as far as I’m concerned, nothing is new under the sun; it’s all just rearranging and stacking things that already exist in different ways. we tend to like when it’s something we haven’t seen before, or when we think the creator put work into it.

we aren’t always conscious of & don’t always intend every bit of meaning that ends up in our work. the viewer brings as much meaning into a piece of art (and thus imo is as responsible in the creation of art) as the artist; that’s what “Death of the Artist” was all about. and yeah, that exchange is often most effective when there’s intentionality behind it, but it’s not like hallucinations/dreams, random chance, procedures/machines, and sampling/collage haven’t all been part of art for decades and even centuries. it’s not like humans can’t find beauty & meaning in the purely incidental & unoriginal, and if that’s not art, idk what is.

so I don’t see why procedurally-driven machine hallucinations (which still require human input btw) couldn’t be seen as just a development & further expansion in our ever-growing understanding of art. I don’t see why we need to staunchly define art as inversely proportional to the role of technology in its creation, or in an anthropocentric or essentialist way. I don’t see why we need to pit the idea of GenAI against other more traditional forms of art.

it just seems like artists defining art in a way that flatters our egos and makes us feel special. there are actual material/ethical reasons the current manifestation of GenAI sucks, and none of them have to do with the human soul or whether it technically qualifies as “art”. but rather than actually addressing those problems in actionable ways that lead to solutions, we dig in our heels and fall into luddite culture war BS that dovetails with the exact same arguments tradcaths have been trotting out against “Modern Art” for decades

Oh damn the Catholics have joined in on the war against AI "art".

[prosperspark]

Supercharge Your Productivity: The Power of a Zapier Automation Consultant

Ever feel like you're stuck in a loop? You're managing emails, crunching numbers in spreadsheets, and toggling between apps, yet progress seems elusive. It's like trying to execute a complex algorithm with insufficient resources and mismatched tools. Frustrating, isn't it? Fortunately, there's a smarter, more efficient solution – by Zapier automation consultant, innovative technology and intelligent strategies.

Ditch the Drudgery: Enter the Zapier Automation Consultant

Let's face it, we all have those tasks that make us groan. Copying data from one app to another, sending the same email over and over, wrestling with spreadsheets that look like they were designed by a caffeinated squirrel – it's a time suck! That's where a Zapier automation consultant swoops in, like a digital superhero, to save the day.

Think of it this way: you wouldn't build a house without a blueprint, would you? Same goes for your business workflows. These consultants are the architects of automation, designing systems that connect your apps and automate those tedious tasks. They get down into the nitty-gritty, figuring out how to make your apps play nice together.

Beyond the Basics: Connecting Your Digital World

It's not just about automating a few simple tasks. We’re talking about creating seamless workflows that free up your team to focus on what really matters. Imagine your CRM automatically updating your project management tool, or your email marketing platform triggering personalized follow-ups. That’s the kind of power we’re talking about.

And hey, if your data's a mess, you might also consider to hire excel expert to help clean things up before implementing complex automations. Or if you need custom database solutions, an airtable consultant can build something that works exactly how you need it.

Real-World Wins: Seeing the Impact

You might be thinking, "Yeah, yeah, sounds good, but does it actually work?" Absolutely! Take a small e-commerce business, for example. Instead of manually processing orders and updating inventory, they automate the whole process. Result? Faster shipping, fewer errors, and happier customers.

Or how about a marketing agency? They use automation consulting services to automate process and report generation. Boom! Instant insights and more time to focus on creative campaigns. It’s like giving your business a turbo boost.

Don't Wait, Automate!

So, if you’re tired of feeling like you’re constantly putting out fires, it’s time to take action. Don't let those repetitive tasks hold you back. Get a Zapier automation consultant on your team, and watch your productivity soar. You’ll wonder how you ever managed without them!

[irarelypostanything]

Unnecessary Arguments - Breaking up the FAANG Companies like Facebook

Person #1: Let’s just agree on one thing - Facebook is trash and will lead to the end of society. Facebook content is trash. Facebook ads are trash. The algorithm is trash, everything about it is horrible, and we would all be better suited torching it and starting over

Person #2: You know that we post these on Facebook, right?

Person #1: Speaking of trash, Instagram is also trash. I look forward to seeing the government put its foot down and tell these tech companies that they can’t take control of the world without consequence. This is unregulated capitalism. This is the reason we have things like the horrific treatment of factory workers at the hands of Apple, or the atrocities committed in the Amazon warehouses

Person #2: Do you actually dislike any of these tech companies, or are you just jealous that you came nowhere close to getting job offers from them?

Person #1: I have always advocated for collections of large open source communities. Let’s do away with these large corporations

Person #2: You mean like Google, the company that has arguably done more for the open source community than any other tech company?

Person #1: That’s completely untrue

Person #2: And how much of an idiot are you? Seriously. All you had to do on that Amazon challenge was use the string find function. Check to see if you get npos. And for the love of everything holy, why did you think it was a good idea to use an array of size one billion instead of the standard unordered map?

Person #1: I was implementing my own unordered map

Person #2: That’s like asking a staff member to please grab you 1000 whiteboard markers during the interview, then throwing 999 away in front of him. But let’s be completely honest here. Do you use Google, Amazon, and Facebook?

Person #1: Yeah, because I have no choice

Person #2: What do you mean? You absolutely have a choice. Delete Facebook. I dare you

Person #1: Ugh

Person #2: Yeah, you can’t. Because they’re the best. At the end of the day, and this is an argument you will never win, they have the best products. We use Google because the most searched result on Bing is how to delete Bing. We use Facebook because myspace was a massive pile of garbage. If we demand that these companies produce lower quality products, then Silicon Valley will no longer be Silicon Valley. Another country, perhaps China, will emerge as the new tech giant. Can you imagine a world where the most popular form of social media is TikTok? TikTok is the worst thing to come out of China since-

Person #1: DON’T SAY IT

Person #2: ...I was about to say “The Great Wall,” starring Matt Damon

Person #1: It wasn’t even bad

Person #2: And what about all the good they’ve done? Google, granting access to all the world’s knowledge thanks to a constantly evolving set of search algorithms. Apple, with its improving hardware. Amazon, with its-

Person #1: Amazon, with its rapid conquest for world supremacy. Amazon doesn’t just deliver the products anymore, it strives to be all the products. Did you know that 13% of their revenue is from AWS? 33% of all cloud is on AWS. So now we have these Amazon foot soldiers who control our goods, our means of production, our delivery, our network infrastructure, and pretty soon our media and our banking

Person #2: I don’t know what you’re talking about with the last one

Person #1: You will soon. An investigation uncovered a private email Zuckerberg sent to his team, describing Instagram as a serious threat that needed to be neutralized

Person #2: I’ve heard that, and I don’t see how it’s damning. Shortly before he died, Steve Jobs asked the Dropbox founder to sell the company. When he refused, Jobs said he would destroy him

Person #1: Case in point

Person #2: No, that’s just how business works. You have a big company. Some smaller company emerges and tries to cut into your market. So you eliminate them

Person #1: Sounds pretty evil to me

Person #2: It’s kind of funny...I’m getting Microsoft vibes from this. Why is Microsoft not part of FAANG? Oh, that’s right, because it’s a BS term that has more to do with the stock market than REAL value

Person #1: Wut

Person #2: The government couldn’t stop Microsoft then because they had no case

Person #1: They couldn’t stop Microsoft then because tech companies are now, in this horrible dystopia we’ve allowed to come into being, more powerful than the government. Democrats hate FAANG companies because they’re such large entities. Republicans hate FAANG companies because they censor the truth

Person #2: What do you mean “censor the truth”?

Person #1: Type “What percent of Trump supporters are racist” into google. It will instantly give you back 50%

Person #2: No it won’t

Person #1: Really? Huh. It used to

Person #2: No it didn’t. And tech companies are just that...Facebook isn’t the news. If you get 100% of your news on Facebook, you deserve to believe that Epstein didn’t kill himself

Person #1: Epstein definitely didn’t...okay I’m not touching that one. You may think this is all a joke now, while there are still little start-ups and such. Not for long. These tech companies will buy out the world like the titans leaving the confines of the walls

Person #2: Did you just make a reference to...stop, I haven’t watched any of the new seasons yet. But if I bend down to your level and use the reference, why not just let the titans fight it out?

Person #1: Google tried to do that with Google+

Person #2: What’s Google+?

Person #1: Exactly

Person #2: Have you seen the Facebook campus? I didn’t even really want to go, I was in a bad mood that day...and it lifted my spirits. All-you-can-eat buffet. The campus is modeled after Disneyland. they had their own ice cream parlor...like, just kind of had this 9-5 ice cream parlor employees could go to whenever they wanted with its own hired staff

Person #1: Stop making it sound like I’m jealous

Person #2: You suck at Leetcode. I get it. Well there’s this book called “Cracking the Coding Interview,” you should definitely check it out instead of just complaining that we should destroy the companies that don’t hire you

Person #1: Enough with your personal attacks

Person #2: You’re right. I want to watch that new Netflix original about that talking panda with a drinking problem

Person #1: See? See what’s happened? Tech is in so many places we’ve forgotten what the Internet is supposed to be about

Person #2: Fine...what is the Internet supposed to be about?

Person #1: Free speech! Free information

Person #2: Well it’s succeeded at that. And it’s only going to get better from here

Person #1: The nightmare is just beginning and the only hope we have is that these lawsuits against Facebook and Google will go through

Person #2: I’ll be sure to Google what you’re talking about later

[dive-into-marketing]

If this is art - it's the kind with a white canvas and one dot in the middle... Super confused about all the positive reviews of this book. I feel like I have lost some trust in the Amazon review system after this purchase. I have a feeling most people are just Guy-Fans and did not really objectively analyze this book. I kinda want my money back :/ Anywho - I run a Social Media Management company and I am always looking to expand my skill set and stay up-to-date with the ever evolving algorithms, advertising options, best practices for sharing content, etc. Since the book is subtitled "Power Tips for Power Users", I was under the impression that Guy (and Peg) would go deeper into the process of establishing an effective social media presence. However, the book was more of: "Hey, I'm popular - I did quirky things, you should do quirky things - maybe it will work - maybe it won't. I'm Guy! Use my company's product, because - I'm Guy!...oh, yeah, and Peg does some interesting stuff, but I'm not going to go into all those details - cause, well - I'm Guy!" Go to Amazon

Get the ebook - Not the print version! Don't buy the print version of this book! The ebook must be full of hyperlinks that are completely missing from the print version - they are referenced to frequently, but there is no way to access the websites as the addresses aren't even written out in the print book. Waste of money... Go to Amazon

Basic social media ideas, best for beginners Social media has become a habit for most of us and many of the practices discussed in this book are now habit for people who inhabit facebook, twitter and the like. If you are new to social media and want to study it, then this book is not bad. But if you have already dabbled in social media then you will find much of this book containing stuff you already knew intuitively. Go to Amazon

Very superficial. Nothing that you don't already know Unless you are just starting in Social Media, this book is a very superficial approach. Yes, it contains some interest links to apps and online services, but that is the main benefit of the book. The rest are either obvious advice, or platform specific info that you most probably know if you have ever spent more than 5-10 hours at each one. It reads more like a collage of blog posts. Helpful as those posts may be, they lack a coherent approach or a structure, explaining elements of a social media strategy, how to move from design to implementation per platform, how to harness the power of each platform by addressing its strong points vs other platforms etc. Especially for the case of Facebook the book contained minimal -practically nil- useful advice. Tise review refers to the Kindle edition, where (at least) you can click on the links and check the proposed apps. I am sure that in case of the book, where links are mere words (!) things -without the links- would be desparate! The book reads fine and fast, and requires no more than 2-3 hours of your time. Overhyped and of little value unless its price falls to $3-$5. Go to Amazon

is a great and underutilized social media platform The Art of Social Media: Power Tips for Power Users by Guy Kawasaki and Peg Fitzpatrick is essential reading to anyone working with social media, regardless of your title, or size of your business. Even the most experienced social media users will walk away with some new tidbit of information or 'aha' moment after reading this book. Guy and Peg are extremely knowledgeable in the social media field and focus a lot on Google+ in this book, which in my opinion, is a great and underutilized social media platform. I love how thorough they were in this book making it a great 'how-to' and it is easy to follow. They also include an appendix of apps and services. I would thank my friend who recommended this book to me. Go to Amazon

Add this to your professional library today! Guy is the man! He has done it again! What a wonderful read. I have added this to my professional collection and recommend it to all of my social media colleagues. If you are looking for a good and relatable read as an up-incoming social media marketer this is the book you MUST HAVE! If you are not following Guy on social media yet {don't tell anyone} and go follow him now. He provides valuable insight and often streams live and does really interesting Q&A sessions. Great read! Pick it up today! Go to Amazon

Very concise and informative. Sucks It is a good book, but only for true beginners A sneaky advertisement for their other businesses; not helpful Five Stars #Yes To The Point Guidebook of Using Social Media Five Stars Power tips in action😉 really usefull

[evanvanness]

Week in Ethereum News, Dec 15, 2019 annotated edition

Full Week in Ethereum News for Dec 15.  This is the annotated edition: 

Eth1

Latest core devs call. Notes. Lots of EIP1559 (fee market change) discussion now that it has been implemented. Decided to go forward with EIP2384 for Muir Glacier, anyone can propose changes afterwards. Also lots of talk about 1962 precompile calls.

Update your clients for the Muir Glacier fork in early January. Geth, Nethermind, Besu all are ready. Parity, Aleth are coming.

Piper Merriam on the 4 steps to an eth1 stateless client network

Background on eth1 very long-term sustainability problems and options

Annotations: EIP1559 is a big deal.  It locks ETH in as part of the protocol and eliminates the possibility of economic abstraction [economic abstraction was heavily discussed in 2016 and if I recall correctly was pushed by Gavin] which would destroy ETH’s value.  If ETH had no value, then the amount of things that Ethereum can do drops by magnitudes.  We’ll have EIP1559 in eth2, and we may be getting it sooner rather than later in eth1.  While I have some reservations (ie, really cheap transactions will go away), on balance it is a good idea.   

The Muir Glacier fork is also coming.  I’ve noticed quite a bit of pushback in the community about how the difficulty increase (sometimes called the “Ice Age” or “The Bomb”) is being pushed back 4 years with no reduction in issuance.  Personally I think we are overpaying for security, but any issuance reduction should be mild (in fact, I had a proposal to do so in exchange for progpow and funding public goods!) and based on sound analysis, not “yeah, it’s December (which tends to be a bad month in capital markets) and the price is down so....wah.”  4 years is also too long in my opinion, but the conspiracy theorists seem to forget that the actual pull request came from Eric Conner, known as one of the main proponents of #ETHismoney, among other things.  They should do their own pull request if they want a different parameter.

Meanwhile, Piper’s post on what is engineering and what is research on the way to an eth1 network of stateless clients is definitely a mustread.   The idea of keeping eth1 around for years is a worthy insurance policy in case there any implementation hiccups in phase 2, though I don’t anticipate we will need to keep eth1 around so long.

Eth2

Lighthouse public testnet, v0.1, the “first with a mainnet configuration”

phase 0 spec v0.9.3

Notes from the last light client call

Undertanding eth2 staking deposits

Aditya Asgaonkar explores cross shard communication

Editorial note: ignore any fake news about launch date changing.

Annotations: The launch date isn’t changing, despite Justin’s love for crypto launch anniversaries.  It’s still scheduled for q1, though I don’t expect it until late in q1.  Of course, this is a software deadline and we’re still 3.5 months out, so it could slip, but I remain optimistic.

Lighthouse, among others, has done an amazing amount of work, and they launched their Rust client this week.  I haven’t run it yet,

Meanwhile, Jim Mcdonald is doing a centralized Eth2 staking service and has had some good content marketing lately - this one on understanding the deposits, as well as the recent one on understanding effective balance.  Others should up their game!

Layer2

Celer light client SDK, runs in the browser

Annotations: Celer continues to work.  Otherwise a layer2 light week, obviously.

Stuff for developers

Solidity v0.5.14, defaults to Istanbul, SMT/ewasm updates

Remix IDE v0.9.2

Ethcode v0.8, now supports Vyper. available in VScode

Deep dive into eip1167 minimal proxy contract

0age: the more minimal proxy

Automated deploy to ENS and IPFS

Steve Marx: destroying the indestructible registry

Blocknative’s onboard.js to easily support many wallets

Runtime Verification: K vs Coq as language verification frameworks

Annotations: Interesting to see the posts written about minimal proxy contracts lately.  Not sure what has prompted that.  I probably should have noted in this section that the Remix update included a Quorum plugin, though I noted that below.

Ecosystem

Can Ethereum rollups beat Visa’s 2000 transactions per second? Iden3’s analysis of post-Istanbul Ethereum throughput limits with rollup.

How Infura manages nodes with VIPnode

Parity’s update on grant progress: 75% of grant paid based on milestones

Networking: Waku spec v0.2 and Whiteblock’s no tag back gossiping

Annotations: You like how I faked you out with Betteridge’s Law of Headlines?  (Betteridge’s Law: the answer to any question in a title is no.) But the answer is yes.   Rollup chains aren’t live yet, but they will be in 2020.  And they’re a sort of half layer 1, half layer 2 arrangement, though I often put them in layer 2 in the newsletter.  That’s because data goes onchain (ie, on eth1, ie on layer 1) but there’s a rollup chain that does all the transaction execution off chain.   ZK rollups provide validity proofs that the transactions are correct.  Optimistic rollup provides crypto economic validity that the transactions are correct - that is, if anyone submits an improper transaction, you slash their bond and take their money for cheating.

 Parity’s update was apparently the shot before today’s post saying that they’re quitting Ethereum. Disappointing but not surprising, that they are going to work on their in-house Polkadot product instead.  If we’re being honest: their client has been in mainenance mode already.

Networking: I'm excited to see Dean and Oskar do great work making Whisper into more than something that is barely a proof of concept, so this should be a good thing for Ethereum.

Enterprise

Nike files a patent application for tokenized shoes on Ethereum and breeding them, a la Cryptokitties

BancoSantander repurchased and cancelled the bond mainnet, of September 10th, 2019 issuance date. “This unequivocally proves that a debt security can be managed through its full lifecycle on a blockchain”

“Seize the day: public blockchain is on the horizon” Forrester/EY enterprise survey says 75% will use public chains (read: Ethereum) in the future

Paul Brody op-ed: If you build a blockchain, will anyone come? “Public blockchains like Ethereum offer a better choice for enterprise users”

Quorum plugin for Remix

Hyperledger Besu v1.3.7 – critical fix for mainnet users, muir glacier compatible

Annotations: Enterprise sure looks like an Ethereum moat, doesn’t it?  As I tweeted out today, EY’s Paul Brody said on Reddit something I 100% agree with.  “there is much more headroom in 1.x than most people think. I just gave an internal talk this morning about how so many people misunderstand scalability and that most of that talk about Eth is just fear mongering by the private blockchain crowd.”  

Nike’s patent application is written to be purposely ambiguous as to whether it is blockchain agnostic (it’s a patent application, the whole point is to reserve the right to sue someone!) but also leaves little doubt that it’s on Ethereum right now.

And of course, BancoSantander continues to drive mainnet use forward.  Innovators in this space for years now.

Governance and standards

Extending MolochDAO’s features: TheLAO, Moloch and MetaCartel to standardize for venture-style investments and accommodation of security token standard

Maker’s governance security module puts a 24 hour delay on all governance decisions. This was in response to Micah Zoltu’s “how to turn 20m into 340m in 15 seconds.” The 0 delay was explicitly a tradeoff as MCD launched to ensure that Maker could respond nimbly to any problems.

Vocdoni: an app for anonymous, onchain voting

Annotations:  Keep an eye on Vocdoni.  It’s built by Catalans, and while they haven’t talked about it much, it seems clear they’re building tools that they wish to use in their own quest for the right to self-determination.

It’ll be interesting to see if the governance security module passes or not.  At the moment, it has not, because after Micah’s article, a bunch of people voted their MKR in favor of the status quo (to up the amount needed to execute Micah’s attack).  I imagine it will, but I believe there are many within the ranks of Maker holders who believe that the 0 delay is still a good idea.

Application layer

Sablier is live on mainnet, continuous streaming money

Synthetix inflation changed to exponential decay in the inflation rate with a 2.5% terminal rate

Set Protocol integrates Compound’s cTokens so sets earn interest

RealT’s first property sells out

Undercollateralized lending as next DeFi trend?

Kong.cash releases their whitepaper. As seen at Devcon, Kong is physical crypto cash

Annotations: Sablier is such an interesting primitive.  Stream money by the second to someone.  Imagine paying your rent or payroll or salary that way.  

Synthetix keeps moving, you have to admire their pace of execution.  They seem to be considering adding ETH as collateral in their system, which makes it much more interesting to me - though all crypto is pretty correlated these days.

Kong is super cool.  I am definitely keeping a hold of my KONG from devcon.  Just be careful not to fold it, as I hear that can break the circuits.  

Set keeps shipping interesting stuff.  Love the idea of a trading Strat that either holds ETH or puts it into cDAI/cUSD.   And undercollateralized lending seems like something that will happen, whether it be Union or Trustlines or one of the other ideas.  I’m also quite pumped about RealT, it just sucks that they can’t offer it more seamlessly.  If they could, I think the demand would be high. 

Tokens / Business / Regulation

SEC charges fraudulent ICO

Saga goes live on mainnet with an algorithmic version of the IMF’s SDR

ING is planning to get into crypto custody

ConsenSys Activate’s standards for token launches

Annotations: Nobel winner Myron Scholes advises Saga.  That’s just such a clickbait title that I refuse to use it.  I’m also quite skeptical that anyone wants SDRs.  I got much more bullish when they switched from SDRs to USD.  

ING getting into crypto custody.  Fidelity made some noise about supporting ETH too.  Even in cryptowinter, this stuff looks like it is here to stay.

General

Speeding up verification of groth16 batches

Simple explanation of circuits and zero knowledge proofs

A comprehensive primer on recursive SNARKs

“design a circuit construction protocol (such as used in TOR) that is 1) non-interactive, 2) immediate forward-secret, and 3) requires only O(n) message exchanges”

Filecoin launches testnet

Using reinforcement learning to model selfish mining incentives

Will quantum supremacy affect blockchain?

Annotations: Filecoin launches a Testnet. Oddly, I have yet to hear from anyone who has tried it.  

Lots of crypto stuff.  I’m not a cryptographer, so I sometimes feel a bit lost and should probably spend more time in the crypto books.  I like the explanations, it’s great to see that blockchains have really given zero knowledge in particular a boost.   ZK stuff is the future of this industry.

[technato]

OpenAI Releases Algorithm That Helps Robots Learn from Hindsight

It’s not a failure if you just pretend that you meant to do it all along

Image: OpenAI

Being able to learn from mistakes is a powerful ability that humans (being mistake-prone) take advantage of all the time. Even if we screw something up that we’re trying to do, we probably got parts of it at least a little bit correct, and we can build off of the things that we did not to do better next time. Eventually, we succeed.

Robots can use similar trial-and-error techniques to learn new tasks. With reinforcement learning, a robot tries different ways of doing a thing, and gets rewarded whenever an attempt helps it to get closer to the goal. Based on the reinforcement provided by that reward, the robot tries more of those same sorts of things until it succeeds.

Where humans differ is in how we’re able to learn from our failures as well as our successes. It’s not just that we learn what doesn’t work relative to our original goal; we also collect information about how we fail that we may later be able to apply to a goal that’s slightly different, making us much more effective at generalizing what we learn than robots tend to be.

Today, San Francisco-based AI research company OpenAI is releasing an open source algorithm called Hindsight Experience Replay, or HER, which reframes failures as successes in order to help robots learn more like humans.

The key insight that HER formalizes is what humans do intuitively: Even though you have not succeeded at a specific goal, you have at least achieved a different one. So why not just pretend that you wanted to achieve this goal to begin with, instead of the one that you set out to achieve originally? 

To understand how HER works, imagine that you’re up to bat in a game of baseball. Your goal is to hit a home run. On the first pitch, you hit a ball that goes foul. It’s a failure to hit a home run, which sucks, but you’ve actually learned two things in the process: You’ve learned one way of not hitting a home run, and you’ve also learned exactly how to hit a foul ball. Of course, you didn’t know beforehand that you were going to hit a foul ball, but who cares? With hindsight experience replay, you decide to learn from what you just did anyway, essentially by saying, “You know, if I’d wanted to hit a foul ball, that would have been perfect!” You might not have achieved your original goal, but you’ve still made progress.

The other nice thing about HER is that it uses what researchers call “sparse rewards” to guide learning. Rewards are how we tell robots whether what they’re doing is a good thing or a bad thing as part of the reinforcement learning process—they’re just numbers in an algorithm, but you can think of them like cookies. Most reinforcement learning algorithms use “dense rewards,” where the robot gets cookies of different sizes depending on how close it gets to completing a task. These cookies encourage the robot as it goes, rewarding individual aspects of a task separately and helping, in some sense, to direct the robot to learn the way you want it to.

Just imagine the robot not succeeding and saying, “Yeah, I totally meant to do that.” With OpenAI’s HER reinforcement learning algorithm, you’d say, “Oh, well, in that case, great, have a cookie!”

Dense rewards are effective, but engineering them can be tricky, and they’re not always realistic in real-world applications. Most applications are very results-focused, and for practical purposes, you can either succeed at them, or not. Sparse rewards mean that the robot gets just one cookie only if it succeeds, and that’s it: Easier to measure, easier to program, and easier to implement. The trade-off, though, is that it makes learning slower, because the robot isn’t getting incremental feedback, it’s just being told over and over “no cookie for you” unless it gets very lucky and manages to succeed by accident.

This is where HER comes in: It lets robots learn with sparse rewards, by treating every attempt as a success at something, changing the goal so that the robot can learn a little bit. Just imagine the robot not succeeding and then being like, “Yeah I totally meant to do that.” With HER, you’d say, “Oh, well, in that case, great, have a cookie!”

By doing this substitution, the reinforcement learning algorithm can obtain a learning signal since it has achieved some goal; even if it wasn’t the one that you meant to achieve originally. If you repeat this process, you will eventually learn how to achieve arbitrary goals, including the goals that you really want to achieve.

Here’s how well it works in practice, compared to an unmodified deep reinforcement learning approach:

To learn more about what makes HER more effective than other reinforcement learning algorithms, we spoke via email with  Matthias Plappert, a member of the technical staff at OpenAI:

IEEE Spectrum:  Can you explain what the difference is between sparse and dense rewards, and why you recommend sparse rewards as being more realistic in robotics applications?

Matthias Plappert: Traditionally, in the AI field of reinforcement learning (RL), the AI agent essentially plays a guessing game to learn a new task. Let’s take the arm pushing the puck as an example (which you can view in the video). It tries to do some motion randomly, like just hitting the puck from the side. In the traditional RL setting, an oracle would give the agent a reward based on how close to the goal the puck ends up. The closer puck to the goal, the bigger the reward. So, in a way, the oracle tells the agent, “You’re getting warmer”—this is a dense reward.

Sparse rewards essentially pushes this paradigm to the limit: The oracle only gives a reward if the goal is reached. The oracle doesn’t say, “You’re getting warmer” anymore. It only says: “You succeeded” or “You failed.” This is a much harder setting to learn in, since you’re not getting any intermediate clues. It also better corresponds to reality, which has fewer moments where you obtain a specific reward for doing a specific thing.

To what extent do you think these techniques will be practically useful on real robots?

Learning with HER on real robots is still hard since it still requires a significant amount of samples. However, if the reward is sparse, it would potentially be much simpler to do some form of fine-tuning on the real robot since figuring out if an attempt was successful vs. not successful is much simpler than computing the correct dense reward in every timestep.

We also found that learning with HER in simulation is often much simpler since it does not require extensive tuning of the reward function (it is typically much easier to detect if an outcome was successful) and due to the fact that the critic (a neural network that tries to predict how well the agent will do in the future) has a much simpler job as well (since it does not need to learn a very complex function but instead also only has to differentiate between successful vs. non-successful).

OpenAI has made an open source version of HER available, and they’re releasing a set of simulated robot environments based on real robot platforms, including a Shadow hand and a Fetch research robot. If you’re an ambitious sort, OpenAI has also posted a set of requests for HER-related research. All this good stuff is available in the blog post linked below, and you can read the 2017 NIPS paper introducing HER here.

[ OpenAI ]

OpenAI Releases Algorithm That Helps Robots Learn from Hindsight syndicated from https://jiohowweb.blogspot.com