Author: azeeadmin

On the heels of a $600 million fundraise earlier this year, payments giant Stripe has been on an acquisition march to continue building out its business. In the latest development, the company has acquired Bouncer, a startup based in Oakland that has built a platform to automatically run card authentications and detect fraud in card-based online transactions. Its technology is tailored for mobile transactions and includes a flow to help users authenticate themselves if they are mistakenly flagged, to come back into an app legitimately (hence the name).

Terms of the deal are not being disclosed, but Stripe is acquiring both Bouncer’s technology and the team, which will be integrated into Stripe Radar. Started in 2018, Radar is Stripe’s AI-based anti-fraud technology toolset, and most of the tech — which is focused around preventing fraudulent transactions on the Stripe platform — has been built in-house up to now. Stripe says that Radar already prevents “hundreds of millions of dollars of fraud for businesses” each year.

“Bouncer is a great tool for modern internet businesses. It allows them to quickly identify stolen cards, while also ensuring legitimate customers can transact without being blocked,” said Simon Arscott, business lead for Stripe Radar, in a statement. “We’re thrilled to welcome the Bouncer team, and their years of experience building payment authentication software for businesses, to Stripe and to enable their technology for Radar users. With the addition of advanced card scanning capabilities, Stripe Radar will be able block more fraud and further increase revenue for millions of businesses around the world who rely on Stripe.”

The deal comes a couple of weeks after Stripe announced the acquisition of TaxJar to bring cloud-based sales tax calculating tools into its payments platform.

Like Stripe itself, Bouncer was incubated at Y Combinator, in its case as part of its Summer 2019 cohort. In addition to YC, it had raised funding from Commerce Ventures and the Pioneer Fund, but had never disclosed how much it had raised in total.

Not to be confused with the Polish marketing technology startup Bouncer, which provides bulk email verification, Oakland Bouncer was co-founded by Will Megson (CEO) and Sam King (chief scientist), who between them have an interesting pedigree when it comes to identity verification, from academia to working at fast-scaling companies in categories that have been some of the biggest adopters of verification technology.

Both previously worked for years at on-demand transportation service Lyft in fraud, identity and payment management. Before that, Megson was at Groupon; and King, in addition to holding a position as an associate professor of computer science at UC Davis, worked at Twitter on account security, founding the fake accounts team.

Groupon is among the customers that Bouncer currently works with, alongside OfferUp, ibotta and Dealerware. Bouncer will keep its current service and customers up post-deal.

Radar is currently sold in a number of tiers ranging from free to 6p per screened transaction, depending on how it is being used (there is a more basic machine learning tier, and an enhanced tier for fraud teams, and the price varies also depending on whether customers are using Stripe’s standard pricing fees or something else). Stripe also offers a chargeback protection service priced at 0.4% per transaction, as well as analytics tools for Radar customers to get an overview of what is going on.

Stripe says that Radar has blocked more than $1 billion in fraudulent transactions since it was launched.

Bouncer is also currently priced at different tiers, ranging from free to $.15/scan for its basic solution, or a custom price for its more tailored services.

Integrating Bouncer’s card scanning and risk technology into the Radar stack will both sweeten the deal for people to buy those services from Stripe, but also make the tools more effective.

As Stripe describes it, when Radar flags a transaction, Bouncer’s card screening and verification technology will kick in as a “dynamic intervention” to confirm whether or not a customer had a legitimate card at the time of the transaction. This is done to help reduce false positives, which are more frequent in high risk transactions (such as those for big-ticket items, or if a person has been making several transactions in quick succession, or other payment activity that just comes up as unusual in systems).

We’ve been in a wave of new authentication technology that includes things like biometrics and other innovations, but Bouncer takes an approach that is less high-tech at the point of ingestion — needing only a phone’s camera and the card that the customer is using. When a transaction is flagged up and sent to Bouncer for verification, Bouncer works by requesting a picture of the payment card (which can be based on any payment card type and can be a low-light picture).

It then runs that through its PCI- and GDPR-compliant system to see if it’s stolen or real. If it’s real, the transaction continues; stolen and the transaction is cancelled. The whole process can take less than a second (not including the time it takes you to take a picture, of course).

For Bouncer, the idea is that Stripe’s machine learning engine will in turn help Bouncer become more effective.

“I’m excited that we’ll be able to scale our advanced card-verification technology across the Stripe network to help businesses grow their revenue while further reducing fraud behind the scenes,” said Will Megson, CEO of Bouncer, in a statement. “The same signals that Radar learns from will make Bouncer more effective, and Bouncer will, in turn, make Radar more effective. We couldn’t be more excited to join the Radar team.”

Stripe has made a number of acquisitions over the years to bring in key pieces of technology, and in one case — when it acquired PayStack in Lagos (another YC alum) — to help Stripe enter and serve merchants in Africa and more emerging markets overall.

At least two of these have been made in aid of bringing on technologists and technology to build out its compliance and authentication tools. In 2016 Stripe quietly acquired Teapot, a Silicon Valley startup that had been working on APIs for identity verification, trust, credit and other tools needed in financial transactions. Its co-founders spent some years at the company before moving on to other things.

In 2019, Stripe acquired a startup out of Ireland called Touchtech to bring in technology to prepare for Strong Customer Authentication regulations in Europe.

The need for better, more sophisticated tools to ensure online transactions are legit is not going anywhere fast. Malicious hacking — and the consequences that has for obtaining personal data that can be used in consumer fraud — continues to be a persistent threat. And in the meantime, e-commerce continues to become an ever-more mainstream activity, widening the pool of consumers and the chances of things going wrong.

There’s a lot of noise out there. The ability to effectively communicate can make or break your launch. It will play a role in determining who wins a new space — you or a competitor.

Most people get that. I get emails every week from companies coming out of stealth mode, wanting to make a splash. Or from a Series B company that’s been around for a while and hopes to improve their branding/messaging/positioning so that a new upstart doesn’t eat their lunch.

How do you make a splash? How do you stay relevant?

Worth noting is that my area of expertise is in the DevOps space and that slant may crop up occasionally. But these five specific tips should be applicable to virtually any startup.

Leverage your founders

This is especially important if you are a small startup that not many people know about. Journalists don’t want to hear opinions from your head of marketing or product — they want to hear from the founders. What problems are they solving? What unique opinions do they have about the market? These are insights that mean the most coming from the people that started the company. So if you don’t have at least one founder that can dedicate time to being the face, then PR is going to be an uphill battle.

That doesn’t mean there isn’t plenty to do to support these efforts. Create a list of all the journalists that have written about your competitors. Read those articles. How can your founder add value to these conversations? Where should you be contributing thought leadership? What are the most interesting perspectives you can offer to those audiences?

This is legwork and research you can do before looping founders into the conversation. Getting your PR going can be like trying to push a broken-down car up the road: If the founders see you exerting effort to get things moving on your own, they’re more likely to get beside you and help.

Here’s an example: It may be unreasonable to ask a founder to sit down and write a 1,000-word thought leadership piece by the end of the week, but they very likely have 20 minutes to chat, especially if you make it clear that the contents of the conversation will make for great thought leadership pieces, social media posts, etc.

The flow looks like:

1. You come up with topic ideas based on research.
2. The founder picks their favorite.
3. You and the founder schedule a 20-minute chat to get their thoughts on paper.
4. You write up the content based on those thoughts.
   
   

Facebook has failed in its bid to prevent its lead EU data protection regulator from pushing ahead with a decision on whether to order suspension of its EU-US data flows.

The Irish High Court has just issued a ruling dismissing the company’s challenge to the Irish Data Protection Commission’s (DPC) procedures.

The case has huge potential operational significance for Facebook which may be forced to store European users’ data locally if it’s ordered to stop taking their information to the U.S. for processing.

Last September Irish data watchdog made a preliminary order warning Facebook it may have to suspend EU-US data flows. Facebook responding by filing for a judicial review and obtaining a stay on the DPC’s procedure. That block is now being unblocked.

We understand the involved parties have been given a few days to read the High Court judgement ahead of another hearing on Thursday — when the court is expected to formally lift Facebook’s stay on the DPC’s investigation (and settle the matter of case costs).

The DPC declined to comment on today’s ruling in any detail — or on the timeline for making a decision on Facebook’s EU-US data flows — but deputy commissioner Graham Doyle told us it “welcomes today’s judgment”.

Its preliminary suspension order last fall followed a landmark judgement by Europe’s top court in the summer — when the CJEU struck down a flagship transatlantic agreement on data flows, on the grounds that US mass surveillance is incompatible with the EU’s data protection regime.

The fall-out from the CJEU’s invalidation of Privacy Shield (as well as an earlier ruling striking down its predecessor Safe Harbor) has been ongoing for years — as companies that rely on shifting EU users’ data to the US for processing have had to scramble to find valid legal alternatives.

While the CJEU did not outright ban data transfers out of the EU, it made it crystal clear that data protection agencies must step in and suspend international data flows if they suspect EU data is at risk. And EU to US data flows were signalled as at clear risk given the court simultaneously struck down Privacy Shield.

The problem for some businesses is that there may simply not be a valid legal alternative. And that’s where things look particularly sticky for Facebook, since its service falls under NSA surveillance via Section 702 of the FISA (which is used to authorize mass surveillance programs like Prism).

So what happens now for Facebook, following the Irish High Court ruling?

As ever in this complex legal saga — which has been going on in various forms since an original 2013 complaint made by European privacy campaigner Max Schrems — there’s still some track left to run.

After this unblocking the DPC will have two enquiries in train: Both the original one, related to Schrems’ complaint, and an own volition enquiry it decided to open last year — when it said it was pausing investigation of Schrems’ original complaint.

Schrems, via his privacy not-for-profit noyb, filed for his own judicial review of the DPC’s proceedings. And the DPC quickly agreed to settle — agreeing in January that it would ‘swiftly’ finalize Schrems’ original complaint. So things were already moving.

The tl;dr of all that is this: The last of the bungs which have been used to delay regulatory action in Ireland over Facebook’s EU-US data flows are finally being extracted — and the DPC must decide on the complaint.

Or, to put it another way, the clock is ticking for Facebook’s EU-US data flows. So expect another wordy blog post from Nick Clegg very soon.

Schrems previously told TechCrunch he expects the DPC to issue a suspension order against Facebook within months — perhaps as soon as this summer (and failing that by fall).

In a statement reacting to the Court ruling today he reiterated that position, saying: “After eight years, the DPC is now required to stop Facebook’s EU-US data transfers, likely before summer. Now we simply have two procedures instead of one.”

When Ireland (finally) decides it won’t mark the end of the regulatory procedures, though.

A decision by the DPC on Facebook’s transfers would need to go to the other EU DPAs for review — and if there’s disagreement there (as seems highly likely, given what’s happened with draft DPC GDPR decisions) it will trigger a further delay (weeks to months) as the European Data Protection Board seeks consensus.

If a majority of EU DPAs can’t agree the Board may itself have to cast a deciding vote. So that could extend the timeline around any suspension order. But an end to the process is, at long last, in sight.

And, well, if a critical mass of domestic pressure is ever going to build for pro-privacy reform of U.S. surveillance laws now looks like a really good time…

“We now expect the DPC to issue a decision to stop Facebook’s data transfers before summer,” added Schrems. “This would require Facebook to store most data from Europe locally, to ensure that Facebook USA does not have access to European data. The other option would be for the US to change its surveillance laws.”

Facebook has been contacted for comment on the Irish High Court ruling. Update: It has now sent this statement:

“Today’s ruling was about the process the IDPC followed. The larger issue of how data can move around the world remains of significant importance to thousands of European and American businesses that connect customers, friends, family and employees across the Atlantic. Like other companies, we have followed European rules and rely on Standard Contractual Clauses, and appropriate data safeguards, to provide a global service and connect people, businesses and charities. We look forward to defending our compliance to the IDPC, as their preliminary decision could be damaging not only to Facebook, but also to users and other businesses.”

When you’re head-down and nose to the grindstone — I’m looking at all you hard-working early-stage startup founders — it’s easy to miss a deadline for an outstanding opportunity. Case in point: competing in Startup Battlefield at TechCrunch Disrupt 2021 in September.

We want every game-changing, innovative startup — from anywhere around the world — to have a shot at massive exposure to investors, media and other influential unicorn-makers. The $100,000 in equity-free prizemoney would be nice, too, right? That’s why we’re extending our application deadline for another full week.

It won’t cost you a thing to apply or to participate, so don’t let this trajectory-changing opportunity slip past you. Apply to Startup Battlefield here before May 27 at 11:59 pm (PT).

The TechCrunch editorial team will vet every application and ultimately choose roughly 20 startups to go head-to-head. Each team receives weeks of free, rigorous coaching from our seasoned Battlefield team. Your pitch, presentation skills and business model will reach new heights of excellence. You’ll also be ready to deftly handle all the questions you’ll receive from our expert VC judges.

Startup Battlefield plays out over several rounds, with the field progressively narrowing. Each time you make the cut, you’ll repeat your pitch-and-answer session to a new set of judges. All that training, prep and focus leads to a final showdown and one last grab for the brass ring. And then it’s up to the judges to decide which stand-out startup wins the championship and that huge check.

While only one startup wins the money and the title, every team that competes benefits from standing in a global spotlight. Sean Huang, co-founder of Matidor, competed in Startup Battlefield at Disrupt 2020. His team was one of the five finalists. Here’s what he said about his experience:

“Going through Startup Battlefield helped us simplify and improve our pitch. It helped us not only with brand messaging, but also to win other pitch competitions after Battlefield. By pitching in the finals, we booked a demo with one of the final panelists.

We received inbound investment interest from 12 Tier-1 investors, and eight potential key clients came to our website for a demo session. We also received an endorsement letter for our Y Combinator application from a fellow Battlefield participant, with whom we formed a great connection.”

You’re head down and focused — that’s why we’re giving you a one-week extension. So, stop, look up and grab this opportunity to take your startup to whole new levels. Get your nose off that grindstone and apply to Startup Battlefield here before May 27 at 11:59 pm (PT).

Is your company interested in sponsoring or exhibiting at Disrupt 2021? Contact our sponsorship sales team by filling out this form.

While the COVID-19 pandemic has had a devastating impact on countless businesses across the globe, the $118 billion gaming industry not only survived, it thrived, with 55% of American consumers turning to gaming for entertainment, stress relief, relaxation and a connection to the outside world amid lockdowns.

This drove a 20% boost in gaming sales globally and created nearly 20,000 jobs in 2020 alone. And it’s not expected to stop any time soon: According to research company IBISWorld, the industry is set to grow again in 2021, adding to the year-over-year growth the industry has seen in the preceding half-decade.

This is great news for the growing gaming industry and especially those looking to score a job at a company developing the next blockbuster. The unemployment rate is already near zero for those with gaming development and design skills, which means there is an unprecedented opportunity to join the field.

Gamesmith, a digital community dedicated to the gaming industry, currently has more than 5,750 open jobs posted on its site, with roles in design, engineering and animation leading the way. In other words, if you never considered a career in the gaming industry — or thought that your skill set wouldn’t translate — this expanding job market needs employees from all types of backgrounds. Chances are one of your interests — besides gaming, of course — can act as a conduit into finding a career.

One career path for those with art skills — particularly with a talent for digital art tools like Autodesk Maya and Adobe Photoshop — is animation. An animator can do any number of jobs at a gaming company or studio, from building immersive landscapes and cities to modeling what a certain character will look like to designing user interfaces and navigational components. There is significant growth here, too: According to a recent study, most sectors of the animation industry are growing 2% to 3% year over year. According to Gamesmith, the average 30-year-old female artist or animator makes a salary of just under $90,000 a year.

If you’re a whiz with words and witty dialogue, then you might consider applying for a job as a writer at a studio. Writers are responsible for writing everything from the profanity-laced shouts heard in the background in the Grand Theft Auto series to the long speeches in epic adventures like The Legend of Zelda: Breath of the Wild. Employers are looking for writers with a flair for crafting stories and understanding characters, so even if you’re not a career writer, highlight the work you’ve done that fits in the mold of what game publishers are looking for.

One of the most important segments of the industry — and one of the fastest-growing — is developing and designing the gaming experiences themselves. The job market for these roles is predicted to grow by 9.3% between 2016 and 2026, according to the New England Institute of Technology, and the breadth of jobs in this wing of the industry range from level designers to lead designer and developers.

Gamesmith calculated that these jobs account for 16% of the available openings, but make up only 5% of all applications. While you will need at least a computer science degree and an understanding of the fundamentals of programming to land one of these jobs, the payoff for the time spent hitting the books is worth it. Gamesmith estimates that the average 28-year-old male engineer earns a salary north of $100,000.

Even if you don’t have any of these skills to help design a game from the ground up, there are still plenty of ways to break into the industry. No matter how good a game is, it will only be a success if people know about it, so the marketing and promotions teams at studios play a crucial role in making sure that consumers purchase the latest release and that it gets written about. If you have great communication skills and can work through people’s problems, companies always need customer service representatives.

Across all sectors of the gaming world, companies are looking to diversify their workforce and move away from an image as a job sector solely populated by white men in their early to mid-30s. Gamesmith research found that currently 74% of the industry’s workforce is male and 64% is white.

But that is changing. While in 2020, only 24% of studios invested moderate resources into diversity initiatives, out of those studios that did invest those resources, 96% reported at least moderately successful results and improvements to company culture. It may seem slow, but there does seem to be a recognition that the gaming industry needs a more diverse workforce as a way not just to bring more equity to their offices, but to make better games in the future and make the industry look more like the people who play games.

“Diversity isn’t a nicety; it’s a necessity if the industry is going to grow, thrive and truly reflect the tens of millions of people who play games every day in this country,” said Jo Twist, the CEO of the U.K.-based gaming trade association Ukie. “A diverse industry that draws on myriad cultures, lifestyles and experiences will lead to more creative and inclusive games that capture the imagination of players and drive our sector forward.”

Between the push to diversify the industry and a slew of new opportunities in the field, the key takeaway is that there are a wide range of possible careers in the industry and, even if you don’t think they do, your skills probably translate into one of the many roles that a gaming company needs to fill. Avid gamers know that you’re not going to beat a game the first time you turn on your console. So hone your skills, build up your experience and continue your quest to land a job in the industry of your dreams.

Last call, y’all. Today’s the final day to keep more money in your pocket and still enjoy three opportunity-packed days at TechCrunch Disrupt 2021 on September 21-23. Our super early-bird price — $99 — disappears tonight, May 14 at 11:59 pm (PT).

Get your click on, avoid the price hike and buy your pass right here, right now. Unless, of course, you enjoy paying more than necessary. Hey, we don’t judge.

Let’s talk about just some of the reasons why TC Disrupt ranks as the must-attend event for anyone within the early-stage startup ecosystem. Connection is key to growth, and you’ll connect with thousands of attendees from around the world.

An all-virtual Disrupt means you can build relationships and collaborate with Silicon Valley movers and shakers no matter where Google Map pins you. And you never know where in the world you might find the perfect partner or eager investor.

Networking at Disrupt is practically an Olympic sport, and now you have two ways to make those vital connections. First up, here’s what Ada Lau, the manager of market development at Hong Kong Science and Technology Parks Corporation, had to say about networking on the fly.

“The virtual platform’s chat feature made it easy to connect with participants. People got creative using it to promote their business, like posting a LinkedIn profile or offering 15-minute time slots to review business pitches. I even saw a product-naming competition. You could find lots of opportunity rolling through the chat area.”

If you want to target people based on shared business goals, you’ll love CrunchMatch, our AI-powered platform. Answer a few quick questions during registration, and you can schedule one-on-one meetings with the people who can help you move forward — pitch investors, demo products, interview tech talent, build important relationships.

Jessica McLean, the director of marketing and communications at Infinite-Compute, shared this about her networking experience.

“CrunchMatch made networking easy. We sent quick introductions, scheduled meetings with investors and other smart people who could add value to our company. A person we connected with at Disrupt is currently helping us with marketing, which is fantastic.”

Beyond networking, you’ll find plenty of possibilities in our expo area. Hundreds of early-stage startups in the Startup Alley ready to showcase their tech products, services and platforms.

You won’t want to miss any of the incredible speakers, interviews, presentations and panel discussions — and thanks to the combination of live-streaming and video-on-demand with your complementary 3-month Extra Crunch membership, you can catch them all at your convenience.

TechCrunch Disrupt 2021 takes place on September 21-23, but today is the last day you can attend all three days for less than $100. Don’t miss out on serious savings. Stop what you’re doing right now and buy your pass your TC Disrupt 2021. The deadline expires tonight, May 14 at 11:59 pm (PT).

Is your company interested in sponsoring or exhibiting at Disrupt 2021? Contact our sponsorship sales team by filling out this form.

After years in the backwaters of venture capital, edtech had a booming 2020. Not only did its products become must-haves after schools around the globe went remote, but investors also poured capital into leading projects. There was even some exit activity, with well-known edtech players like Coursera going public earlier this year.

But despite a rush of private capital — which has continued into this year, as we’ll demonstrate — edtech stocks have taken a hammering in recent weeks. So while venture capitalists and other startup investors are pumping more capital into the space in hopes of future outsize returns, the stock market is signaling that things might be heading in the other direction.

Who’s right? One investor that The Exchange spoke to noted that market turbulence is just that, and that he’s tuning into activity but not yet changing his investment strategy. At the same time, the recent volatility is worth tracking in case it’s a preview of edtech’s slowdown.

The Exchange explores startups, markets and money. 

Read it every morning on Extra Crunch or get The Exchange newsletter every Saturday.

Let’s look at the changing value of edtech stocks in recent months, parse some preliminary data via PitchBook that provides a good feel for the directional momentum of edtech venture capital, and try to see if there’s irrational exuberance among private investors.

You could argue that it’s public investors who are suffering from irrational pessimism and that private-market investors have the right to it. But since public markets price private markets, we tend to listen to them. Let’s go!

Falling shares

We’re sure that you want to get into the private-market data, so we’ll be brief in describing the public-market carnage. What follows is a digest of edtech stocks and their declines from recent highs:

• Compared to its 52-week high, Chegg stock has lost over a third of its value.
• After reaching $62.53 per share in April, Coursera has shed about half of its value and is trading close to its $33 IPO price.
• 2U closed at $33.92 per share yesterday, its shares also losing half of their value compared to their 52-week high.
• Staying on that theme, Stride (K12) closed at $26.77 per share yesterday, which is about half of its 52-week high.
  
  

Cisco has been busy on the acquisition front this week, and today the company announced it was buying threat assessment platform Kenna Security, the third company it has purchased this week. The two companies did not disclose the purchase price.

With Kenna, Cisco gets a startup that uses machine learning to sort through the massive pile of threat data that comes into a security system on a daily basis and prioritizes the threats most likely to do the most damage. That could be a very useful tool these days when threats abound and it’s not always easy to know where to put your limited security resources. Cisco plans to take that technology and integrate into its SecureX platform.

Gee Rittenhouse, senior vice president and general manager of Cisco’s Security Business Group wrote in a blog post announcing the deal with Kenna, that his company is getting a product that brings together Cisco’s existing threat management capabilities with Kenna’s risk-based vulnerability management skills.

“That is why we are pleased to announce our intent to acquire Kenna Security, Inc., a recognized leader in risk-based vulnerability prioritization with over 14 million assets protected and over 12.7 billion managed vulnerabilities. Using data science and real-world threat intelligence, it has a proven ability to bring data in from a multi-vendor environment and provide a comprehensive view of IT vulnerability risk,” Rittenhouse wrote in the blog post.

The security sphere has been complex for a long time, but with employees moving to work from home because of COVID, it became even more pronounced in the last year. In a world where the threat landscape changes quickly, having a tool that prioritizes what to look at first in its arsenal could be very useful.

Kenna Security CEO Karim Toubba gave a typical executive argument for being acquired: it gives him a much bigger market under Cisco than his company could have built alone.

“Now is our opportunity to change the industry: once the acquisition is complete, we will be one step closer to delivering Kenna’s pioneering Risk-Based Vulnerability Management (RBVM) platform to the more than 7,000 customers using Cisco SecureX today. This single action exponentially increases the impact Kenna’s technology will have on the way the world secures networks, endpoints and infrastructures.,” he wrote in the company blog.

The company, which launched in 2010, claims to be the pioneer in the RBVM space. It raised over $98 million on a $320 million post-money valuation, according to Pitchbook data. Customers include HSBC, Royal Bank of Canada, Mattel and Quest Diagnostics.

For those customers, the product will cease to be stand-alone at some point as the companies work together to integrate Kenna technology into the SecureX platform. When that is complete, the stand-alone customers will have to purchase the Cisco solution to continue using the Kenna tech.

Cisco has had a busy week on the acquisition front. It announced its intent to acquire Sedona Systems on Tuesday, Socio Labs on Wednesday and this announcement today. That’s a lot of activity for any company in a single week. The deal is expected to close in Cisco Q4 FY 2021. The company’s 170 employees will be joining the Security Business Group led by Rittenhouse.

Hello and welcome back to Equity, TechCrunch’s venture capital-focused podcast, where we unpack the numbers behind the headlines.

The fully-vaxxed and officially fully-immune took over the podcast this week, with Natasha and Danny co-hosting the show while the inimitable Alex is out from Shot #2. Grace and Chris, as always, were behind the scenes making sure we sound pretty and don’t fall down too many punny board game rabbit holes after vacation.

Here’s the rundown of what we got into:

• • Crypto crashed –  thanks to either Elon or the guy who created Ethereum and then donated some to relief efforts for COVID-19.  We debated how signal gets lost and rediscovered and lost again in this news cycle, and what that means for anyone in tech trying to get a pulse of. what’s actually happening.
  • One symptom of market craze and disconnected booms? A startup named Caplight that wants more people to buy and sell short positions on private startups. 
  • We then pivoted into the last bit of our ‘The World Is Melting’ section to talk about home ownership and the slow stink of real estate stocks right now. Ontop of that, Better.com, a digital mortgage lender, decided to SPAC after reportedly tabling an IPO.
  • A mention of SPACs obviously opened up a can of worms, or should I say, oasis of woods.
  • The world might be melting, but Softbank could be having the last word. The Japanese conglomerate shut us up with its recent profits – proving Danny’s point in an old EC piece (use code EQUITY for a discount).
  • We switched gears and talked about a neat new $20 million fund coming out of the Midwest, Sixty8 Capital, and its plan to invest in underrepresented founders.
  • The show ended as chaotically focused as it began with a round robin of the most interesting funding rounds of the last week, not limited to but including Blind’s $37 million raise, The Last Gameboard’s $4 million check, Just Women’s Sports, and Morressier’s $18 million Series A.   

And that’s where we break! Follow the podcast on Twitter, be kind to your humans, and be the kindest to yourself. Back sooner than you can raise a $25 million pre-seed round for an audio app for Dogecoin lovers.

Equity drops every Monday at 7:00 a.m. PST, Wednesday, and Friday at 6:00 AM PST, so subscribe to us on Apple Podcasts, Overcast, Spotify and all the casts!

While modern machine learning systems act with a semblance of artificial intelligence, the truth is they don’t “understand” any of the data they work with — which in turn means they tend to store even trivial items forever. Facebook researchers have proposed structured forgetfulness as a way for AI to clear the decks a bit, improving their performance and inching that much closer to how a human mind works.

The researchers describe the problem by explaining how humans and AI agents might approach a similar problem.

Say there are ten doors of various colors. You’re asked to go through the yellow one, you do so and then a few minutes later have forgotten the colors of the other doors — because it was never important that two were red, one plaid, two walnut, etc, only that they weren’t yellow and that the one you chose was. Your brain discarded that information almost immediately.

But an AI might very well have kept the colors and locations of the other nine doors in its memory. That’s because it doesn’t understand the problem or the data intuitively — so it keeps all the information it used to make its decision.

This isn’t an issue when you’re talking about relatively small amounts of data, but machine learning algorithms, especially during training, now routinely handle millions of data points and ingest terabytes of imagery or language. And because they’re built to constantly compare new data with their accrued knowledge, failing to forget unimportant things means they’re bogged down by constant references to pointless or outdated data points.

The solution hit upon by Facebook researchers is essentially — and wouldn’t we all like to have this ability — to tell itself how long it needs to remember a piece of data when it evaluates it to begin with.

Image Credits: Facebook

“Each individual memory is associated with a predicted expiration date, and the scale of the memory depends on the task,” explained Angela Fan, a Facebook AI researcher who worked on the Expire-Span paper. “The amount of time memories are held depends on the needs of the task—it can be for a few steps or until the task is complete.”

So in the case of the doors, the colors of the non-yellow doors are plenty important until you find the yellow one. At that point it’s safe to forget the rest, though of course depending on how many other doors need to be checked, the memory could be held for various amounts of time. (A more realistic example might be forgetting faces that aren’t the one the system is looking for, once it finds it.)

Analyzing a long piece of text, the memory of certain words or phrases might matter until the end of a sentence, a paragraph, or longer — it depends on whether the agent is trying to determine who’s speaking, what chapter the sentence belongs to, or what genre the story is.

This improves performance because at the end, there’s simply less information for the model to sort through. Because the system doesn’t know whether the other doors might be important, that information is kept ready at hand, increasing the size and decreasing the speed of the model.

Fan said the models trained using Expire-Span performed better and were more efficient, taking up less memory and compute time. That’s important during training and testing, which can take up thousands of hours of processing, meaning even a small improvement is considerable, but also at the end user level, where the same task takes less power and happens faster. Suddenly performing an operation on a photo makes sense to do live rather than after the fact.

Though being able to forget does in some ways bring AI processes closer to human cognition, it’s still nowhere near the intuitive and subtle ways our minds operate. Of course, being able to pick what to remember and how long is a major advantage over those of us for whom those parameters are chosen seemingly randomly.