Creating unbiased, accurate algorithms isn’t impossible — it’s just time consuming.
“It actually is mathematically possible,” facial recognition startup Kairos CEO Brian Brackeen told me on a panel at TechCrunch Disrupt SF.
Algorithms are sets of rules that computers follow in order to solve problems and make decisions about a particular course of action. Whether it’s the type of information we receive, the information people see about us, the jobs we get hired to do, the credit cards we get approved for, and, down the road, the driverless cars that either see us or don’t, algorithms are increasingly becoming a big part of our lives. But there is an inherent problem with algorithms that begins at the most base level and persists throughout its adaption: human bias that is baked into these machine-based decision-makers.
Creating unbiased algorithms is a matter of having enough accurate data. It’s not about just having enough “pale males” in the model, but about having enough images of people from various racial backgrounds, genders, abilities, heights, weights and so forth.
“In our world, facial recognition is all about human biases, right?” Brackeen said. “And so you think about AI, it’s learning, it’s like a child and you teach it things and then it learns more and more. What we call right down the middle, right down the fair way is ‘pale males.’ It’s very, very good. Very, very good at identifying somebody who meets that classification.”
But the further you get from pale males — adding women, people from different ethnicities, and so forth — “the harder it is for AI systems to get it right, or at least the confidence to get it right,” Brackeen said.
Still, there are cons to even a one hundred percent accurate model. On the pro side, a good facial recognition use case for a completely accurate algorithm would be in a convention center, where you use the system to quickly identity and verify people are who they say they are. That’s one type of use case Kairos, which works with corporate businesses around authentication, addresses.
“So if we’re wrong, at worst case, maybe you have to do a transfer again to your bank account,” he said. “If we’re wrong, maybe you don’t see a picture accrued during a cruise liner. But when the government is wrong about facial recognition, and someone’s life or liberty is at stake, they can be putting you in a lineup that you shouldn’t be in. They could be saying that this person is a criminal when they’re not.”
But in the case of law enforcement, no matter how accurate and unbiased these algorithms are, facial recognition software has no business in law enforcement, Brackeen said. That’s because of the potential for unlawful, excessive surveillance of citizens.
Given the government already has our passport photos and identification photos, “they could put a camera on Main Street and know every single person driving by,” Brackeen said.
And that’s a real possibility. In the last month, Brackeen said Kairos turned down a government request from Homeland Security, seeking facial recognition software for people behind moving cars.
“For us, that’s completely unacceptable,” Brackeen said.
Another issue with 100 percent perfect mathematical predictions is that it comes down to what the model is predicting, Human Rights Data Analysis Group lead statistician Kristian Lum said on the panel.
“Usually, the thing you’re trying to predict in a lot of these cases is something like rearrest,” Lum said. “So even if we are perfectly able to predict that, we’re still left with the problem that the human or systemic or institutional biases are generating biased arrests. And so, you still have to contextualize even your 100 percent accuracy with is the data really measuring what you think it’s measuring? Is the data itself generated by a fair process?”
HRDAG Director of Research Patrick Ball, in agreement with Lum, argued that it’s perhaps more practical to move it away from bias at the individual level and instead call it bias at the institutional or structural level. If a police department, for example, is convinced it needs to police one neighborhood more than another, it’s not as relevant if that officer is a racist individual, he said.
“What’s relevant is that the police department has made an institutional decision to over-police that neighborhood, thereby generating more police interactions in that neighborhood, thereby making people with that ZIP code more likely to be classified as dangerous if they are classified by risk assessment algorithms,” Ball said.
And even if the police were to have perfect information about every crime committed, in order to build a fair machine learning system, “we would need to live in a society of perfect surveillance so that there is absolute police knowledge about every single crime so that nothing is excluded,” he said. “So that there would be no bias. Let me suggest to you that that’s way worse even than a bunch of crimes going free. So maybe we should just work on reforming police practice and forget about all of the machine learning distractions because they’re really making things worse, not better.”
He added, “For fair predictions, you first need a fair criminal justice system. And we have a ways to go.”
VC firms haven’t been the only ones raising hundreds of millions of dollars to invest in a booming market. After 15+ years of being the last industry anyone wanted to invest in, the music industry is coming back, and money is flooding in to buy up the rights to popular songs.
As paid streaming subscriptions get mainstream adoption, the big music streaming services – namely Spotify, Apple Music, and Tencent Music, but also Pandora, Amazon Music, YouTube Music, Deezer, and others – have entered their prime. There are now over 51 million paid subscription accounts among music streaming services in the US. The music industry grew 8% last year globally to $17.3 billion, driven by a 41% increase in streaming revenue and 45% increase in paid streaming revenue.
The surge in music streaming means a surge in income for those who own the copyrights to songs, and the growth of entertainment in emerging markets, growing use in digital videos, and potential use of music in new content formats like VR only expand this further. Unsurprisingly, private equity firms, family offices, corporates, and pension funds want a piece of the action.
There are two general types of copyrights for a song: the publishing rights and the master rights. The musical composition of a song – the lyrics, melodies, etc. – comes from songwriters who own the publishing right (though generally they sign a publishing deal and their publisher gets ownership of it in addition to half the royalties). Meanwhile, the version of a song being performed comes from the recording artist who owns the master right (though usually they sign a record deal and the record label gets ownership of the masters and most of the royalties).
Popular songs are valuable to own because of all the royalties they collect: whenever the song is played on a streaming service, downloaded from iTunes, or covered on YouTube (a mechanical license), played over radio or in a grocery store (a performance license), played as soundtrack over a movie or TV show (a sync license), and for other uses. More royalty income from a song goes to the master owner since they took on more financial risk marketing it, but publishers collect royalties from some channels that master owners don’t (like radio play, for instance).
For a songwriter behind popular songs, these royalties form a predictable revenue stream that can amount to tens of thousands, hundreds of thousands, or even millions of dollars per year. Of course, most songs that are written or recorded don’t make any money: creating a track that breaks out in a crowded industry is hard. This scarcity – there are only so many thousands of popular musicians and a limited number of legendary artists whose music stays relevant for decades – means copyrights for successful musicians command a premium when they or their publisher decide to sell them.
In 2017, revenue from streaming services accounted for 38% of worldwide music industry revenue, finally overtaking revenue from traditional album sales and song downloads. Subscription streaming services hit a pivot point in gaining mainstream adoption, but they still have far to go. Goldman Sachs media sector analyst Lisa Yang predicted that by 2030, the global music industry will reach $41 billion in market size as the global streaming market multiplies in size to $34 billion (nearly all of it from paid subscriptions).
Earlier this week, I spoke with Merck Mercuriadis who has managed icons like Elton John, Guns N’ Roses, and Beyoncé and raised £200 million ($260 million) on the London Stock Exchange in June for an investment vehicle (Hipgnosis Songs) to acquire the catalogues of top songwriters. His plan is to raise and invest £1 billion over the next three to five years, arguing that the shift to passive consumers paying for music will take the industry to heights it has never seen before.
Indeed, streaming music is a paradigm shift from the past. With all the world’s music available in one interface for free (with ads) or for an affordable subscription (without ads), consumers no longer have to actively choose which specific songs to buy (or even which to download illegally).
With it all in front of them and all included in the price, people are listening to a broader range of music: they’re exploring more genres, discovering more musicians who aren’t stars on traditional radio, and going back to music from past decades. Consumers who weren’t previously buying a lot of music are now subscribing for $120 per year and spreading it across more artists.
Retail businesses are doing the same: through streaming offerings like Soundtrack Your Brand (which spun out of Spotify), they’re using commercial licenses – which are more expensive – to stream a broader array of music in stores rather than putting on the radio or playing the same few CDs.
Much of the music industry’s market growth is happening in China, India, Latin America, and emerging markets like Nigeria where subscription apps are replacing widespread music piracy or non-consumption. Tencent Music Entertainment, whose three streaming services have roughly 75% market share in China (a music market that expanded by 34% last year), is preparing for an IPO that could give it roughly the same $29 billion valuation Spotify received in its IPO in April. Meanwhile, music industry revenue from Latin America grew 18% last year.
Western music is infused in pop culture worldwide, so as these countries enter the streaming era they are monetizing hundreds of millions of additional listeners, through ad revenue at a minimum but increasingly through paid subscriptions as well.
At the talent management, publishing, and production firm Primary Wave, founder Larry Mestel is seeing emerging markets drive more revenue to his clients (like Smokey Robinson, Alice Cooper, Melissa Etheridge, and the estate of Bob Marley) as new fan bases engage with their music online. He raised a new $300 million fund (backed by Blackrock and other institutions) in 2016 to acquire rights in music catalogues amid a market he says has improved substantially due to growth opportunities stemming from the streaming model.
It’s not just streaming music platforms that are driving growth either. Streaming video has exploded, whether it’s from short YouTube videos or the growing number of shows on platforms like Hulu and Amazon Prime Video, and with that comes growing sync licensing of songs for their soundtracks; global sync licensing revenue was up 10% year-over-year in 2017 alone. Over the last year, Facebook signed licenses with every large publisher to cover use of song clips by its users in Instagram Stories and Facebook videos as well.
Catalogues are commonly valued based on the “net publisher’s share,” which is the average amount of annual royalty money left over after paying out any percentages owed to others (like a partial stake in the royalties still held by the artist).
When Round Hill Music acquired Carlin for $245 million in January to gain ownership in the catalogues of Elvis Presley, James Brown, AC/DC, and others, it paid a 16x multiple on net publisher share, which is high but not uncommon in the current market when trading catalogues of legendary artists. Just three years ago, multiples anchored in the 10-12 range (or less for newer or smaller artists whose music has not yet shown the same longevity).
Kobalt, which raised $205 million from VC firms like GV and Balderton Capital to become a technology-centric publisher and label services powerhouse, has also become an active player in the space. Aside from its core operating business (where it stands out from traditional publishers and labels for not taking control of clients’ copyrights), it has raised two funds ($600M for the most recent one) to help institutional investors like the Railpen pension fund in the UK gain exposure to music copyrights as an asset class. In December, their fund acquired the catalogue of publisher SONGS Music Publishing for a reported $160M in a sale process against 13 other bidders looking to buy ownership in songs by Lorde, The Weeknd, and other young pop and hip-hop artists.
The natural question to ask when there’s a rapid surge of money (and a corresponding surge in prices) in an asset class is whether there’s a bubble. After all, last year’s industry revenues were still only 68% of those in 1999 and the rate of growth will inevitably slow once streaming has captured the early majority of consumers.
But the fundamentals driving this capital are in line with a secular shift – it’s evident that music streaming still has a lot of room to grow in a few short years, especially as a large portion of the human population is just coming online (and doing so over mobile first). Plus as new content formats like augmented and virtual reality come to fruition, new categories of music sync licensing will inevitably accompany them for their soundtracks.
Each catalogue is its own case, of course. As Shamrock Capital managing director Jason Sklar emphasized to me, the rising tide isn’t lifting all boats equally. The streaming revolution appears to be disproportionately benefiting hip-hop, rap, and pop given the youth skew of streaming service users and the digital-native social media engagement of the artists in those genres.
Beyond the purchase price, the critical variable for evaluating a deal in this market is also the operational value a potential buyer can provide to the catalogue: their ability to actively promote songs from the past by pitching them to new TV shows, ad campaigns, and any number of other projects that will keep them culturally relevant. This is where strategic investors have an advantage over purely financial investors in publishing rights, especially when it comes to the longer tail of middle-tier artist’s whose music doesn’t naturally get the inbound demand that the Beatles or Prince catalogues do.
With strong long-term market growth and a wide range of possible niches and strategies, music copyrights are an asset class where we’ll see a number of major new players develop.
We cared about Cambridge Analytica because it could have helped elect Trump. We ignored LocationSmart because even the though the company was selling and exposing the real-time GPS coordinates of our phones, it was never clear exactly if or how that data was misused.
The social network’s engineering was sloppy, allowing three bugs to be combined to steal the access tokens of 50 million people. In pursuit of rapid growth at affordable efficiency, Facebook failed to protect its users. This assessment doesn’t discount that. Facebook screwed up big time.
But despite the potential that those access tokens could have let the attackers take over user accounts, act as them, and scrape their personal info, it’s unclear how much users really care. That’s because for now, Facebook and it’s watchdogs aren’t sure exactly what data was stolen or how it was wrongly used.
This could all change tomorrow. If Facebook discovers the hack was perpetrated by a foreign government to interfere with elections, by criminals to bypass identity theft security checkpoints and steal people’s bank accounts or social media profiles, or to target individuals for physical harm, out will come the pitchforks and torches.
Given a sufficiently scary application for the data, the breach could finish the job of destroying Facebook’s brand. If users start clearing their profile data, reducing their feed browsing, and ceasing to share, the breach could have significant financial and network effect consequences for Facebook. After years of scandals, this could be the hack that’s broke the camel’s back.
Yet in the absence of that evil utilization of the hacked data, the breach could fade into the background for users. Similar to the tension-filled departures of the founders of Facebook’s acquisitions Instagram and WhatsApp, the brunt of the backlash may not come from the public.
The hack could hasten regulation of social media. Senator Warner called on Congress to “step up” following the hack. He’s previously advocated for privacy laws similar to Europe’s GDPR. That includes data portability and interoperability rules that could make it easier to switch social networks. That threat of people moving to competing apps could succeed in compelling Facebook to treat user privacy and security better.
One of the biggest questions about the attack is whether the tokens were used to access other services like Airbnb or Spotify that rely on Facebook Login. The breach could steer potential partners away from building atop Facebook’s identity platform. But at least you don’t have to worry about changing all your passwords. Unlike hacks that steal usernames and passwords, the lasting danger of the Facebook breach is limited. The access tokens have already been invalidated, whereas password reuse can lead people to have their other apps hacked long after the initial breach.
If government investigators, journalists, or anti-Facebook activists want to make the company pay for its negligence, they’ll need to connect it to some concrete threat to how we live or what we believe.
For now, without a nefarious application of the breached data, this scandal could blend into the rest of Facebook’s troubles. Every week, sometimes multiple times a week, Facebook has some headline grabbing problem. Over time, those are adding up to deter usage of Facebook and spur more users to delete it. But without an independent general purpose social network they can easily switch to, many users have endured Facebook’s stumbles in exchange for the connective utility it provides.
As breaches become more common, the public may be desensitized. Between Equifax, Yahoo, and the cell phone companies, we’re growing accustomed to letting out a deep sigh with maybe some expletives, and moving on with our lives. The ones we’ll remember will be those where the danger metastasized from the digital world into our offline lives.
[Featured image via Getty]
French startup Klaxit connects drivers with riders so that you don’t have to take your car to work every day. And the company recently announced a new feature with the help of Uber. If your driver cancels your ride home, Klaxit will book an Uber for you.
Klaxit is a ride-sharing startup that focuses on one thing — commuting to work. And this problem is more complicated than you might think. You can’t just go to work with the same person every day because you don’t always go to work at the same time. Similarly, sometimes your driver has to leave work early, leaving you at the office with no alternative.
As a driver, you want to take the quickest route to work. So you want to be matched with riders who are exactly on the way to work.
Klaxit currently handles 300,000 rides per day. In particular, the company has partnered with 150 companies, including big French companies such as BNP Paribas, Veolia, Vinci and Sodexo.
Klaxit can be particularly useful for companies with large office buildings outside of big cities. Promoting Klaxit instantly fosters supply and demand from and to this office. But you don’t have to work for one of those companies to use Klaxit.
Local governments can also financially support Klaxit to improve traffic conditions and mobility for users who don’t have a car or a driver’s license. “Subsidizing rides on Klaxit is 8 to 10 times cheaper than building a bus line,” co-founder and CEO Julien Honnart told me.
One of the biggest concerns as a rider is that you’re going to be stuck at work in the evening. Klaxit is now asking its users to request a ride with two other drivers. If they both decline your request, Klaxit will book you an Uber ride to go back home.
You don’t have to pay the Uber ride and then get reimbursed, Klaxit pays Uber directly. You don’t need an Uber account either as Klaxit is using Uber for Business. MAIF is the insurance company behind this insurance feature, and also one of Klaxit’s investors. This is a neat feature to convince new users that they can trust Klaxit.
French startup Ownpage has recently released a new product called Relike. Relike is one of the easiest ways to get started with email newsletters. You enter the web address of your Facebook page and that’s about it.
The company automatically pulls your most recent posts from your Facebook page and lets you set up an emailing campaign in a few clicks. You can either automatically pick your most popular Facebook posts or manually select a few posts.
Just like any emailing service, you can choose between multiple templates, decide the day of the week and time of the day, import a database of email addresses and more. If you’ve used Mailchimp in the past, you’ll feel right at home.
But the idea isn’t to compete directly with newsletter services. Many social media managers, media organizations, small companies, nonprofits and sports teams already have a Facebook page but aren’t doing anything on the email front.
Relike is free if you send less than 2,000 emails per month and don’t need advanced features. If you want to get open rates, click-through rates and other features, you’ll need to pay €5 per month and €0.50 every time you send 1,000 emails.
The company’s other product Ownpage is a bit different. Ownpage has been working with media organizations to optimize their email newsletters. The company is tracking reading habits on a news site and sending personalized email newsletters.
This way, readers will get tailored news and will more likely come back to your site. Many big French news sites use Ownpage for their newsletters, such as Les Echos, L’Express, 20 Minutes, BFM TV, Le Parisien, etc.
Ownpage founder and CEO Stéphane Cambon told me that Relike was the obvious second act. Using browsing data for customized newsletters is one thing, but many talented social media managers know how to contextualize stories and maximize clicks (even if it means clickbait, sure).
The startup was looking at a way to get this data, and ended up creating Relike, which could appeal to customers beyond news organizations. For now, both products will stick around. In the future, the company plans to add Twitter and Instagram integrations as well as better signup flows for newsletter subscribers.
I looked at the TIOBE index today, as I do every so often, as most of the software pros I know do every so often. It purports to measure the popularity of the world’s programming languages, and its popularity-over-time chart tells a simple story: Java and C are, and have been since time immemorial, by some distance the co-kings of language.
But wait. Not so fast. The rival “PYPL Index” (PopularitY of Programming Languages) says that Python and Java are co-kings, and C (which is lumped in with C++, surprisingly) is way down the list. What’s going on here?
What’s going on is that the two indexes have very different methodologies … although what their methodologies have in common is both are very questionable, if the objective is to measure the popularity of programming languages. TIOBE measures the sheer quantity of search engine hits. PYPL measures how often language tutorials are Googled.
Both are bad measures. We can expect the availability of online resources to be an extremely lagging indicator; a once-dominant dead language would probably still have millions of relict web pages devoted to it, zombie sites and blog posts unread for years. And the frequency of tutorial searches will be very heavily biased towards languages taught en masse to students. That’s not a meaningful measure of which languages are actually in use by practitioners.
There are lots of weird anomalies when you look harder at the numbers. According to TIOBE, last C went from its all-time lowest rating to Programming Language Of The Year in five months. I can buy that C has had a resurgence in embedded systems. But I can also easily envision this being an artifact of a highly imperfect measure.
The more flagrant anomaly, though, in both of those measures, is the relative performance of Objective-C and Swift, the two languages used to write native iOS apps. I can certainly believe that, combined, they have recently seen a decline in the face of the popularity of cross-platform alternatives such as Xamarin and React Native. But I have a lot of trouble believing that, after four years of Apple pushing Swift — to my mind, an objectively far superior language — Objective-C is still more popular / widely used. In my day job I deal with a lot of iOS/tvOS/watchOS apps, and interview a lot of iOS developers. It’s extremely rare to find someone who hasn’t already moved from Objective-C to Swift.
But hey, anecdotes are not data, right? If the only available measures conflict with my own personal experience, I should probably conclude that the latter is tainted by selection bias. And I’d be perfectly willing to do that …
… except there is another measure of programming language popularity out there. I’m referring to GitHub’s annual reports of the fifteen most popular programming languages on its platform. Those numbers are basically a perfect match for my own experience … and they are way disjoint from the claims of both both TIOBE and PYPL.
Obviously the GitHub numbers are not representative of the entire field either; their sample size is very large, but only considers open-source projects. But I note that GitHub is the only measure which counts Swift as more popular than Objective-C. That makes it a lot more convincing, to me … but its open-source selection bias means it’s still far from definitive.
These statistics do actually matter, beyond being an entertaining curiosity and/or snapshot of the industry. Languages aren’t all-important, but they’re not irrelevant either. People determine what languages to study, and sometimes even what jobs to seek and accept, based on their popularity and their (related) projected future value. So it’s a little upsetting that these three measures are so starkly, radically different. Sadly, though, we seem to still be stuck with tea leaves rather than hard numbers.
Since McKinsey released a report on how best to use prizes to incentivize innovation nearly a decade ago, an entire industry has grown around social innovation challenges. The formula for these “save the world” competitions has become standard. Drum up a lot of buzz around an award. Partner with big names to get funding and high-profile judges. Try and get as many submissions as possible from across the world. Whittle down the submissions and come up with a list of finalists that get to pitch at a glitzy event with a lot of media attention.
On the final stage, based on pitches that last for mere minutes, judges typically pick one winner that can get upwards of millions in prize funding. Don’t have a software platform to run a challenge of this kind? No worries, numerous for-profit vendors have sprung up that can do all the work for you—for anywhere from ten to a few hundred thousand dollars. The growth has been so exponential that prizes awarded through competitions has grown from less than $20 million in 1970 to a whopping $375 million just four decades later.
But do these prizes get the sort of world-saving results they aim for? There’s little quantified evidence to back that, and some leaders in philanthropy are broadly skeptical.
For its part, the Massachusetts Institute of Technology is trying a different approach to innovation challenges with Solve, taking some of what’s worked in these challenges and fusing it with elements of tech accelerator programs, including a post-award training program that focuses on results.
Solve is entering an already crowded field of innovation challenges. Many of these prizes overlap, with each vying to be the “Nobel” of its field. More prizes means more noise—which has led to a race to offer more money to get attention.
But even private-sector riches do not guarantee that prize money for innovation gets good results. In 2004, Bigelow Enterprises sponsored a $50 million Space Prize but it failed to capture the imagination of space researchers and eventually folded. Back in 2009, Netflix invited outside teams to improve it movie recommendation algorithm by 10% for a $1 million reward. The Netflix Prize led to a race among programmers, only for Netflix to eventually kill the entire plan because it was getting better results in-house.
Overall, the social innovation competitions tend to reward presentation, glitz and charisma, and penalize speaking English as a second language, introversion and inability to make flashy slides.
Now let’s take a look at Solve, which held its third annual finalists event on Sunday September 23 in New York.
Unlike other contests where questions are internally decided, Solve crowdsources the questions to begin with. Its team takes months to run hackathons and workshops around the world to decide on the four most pressing questions to become the focus of that year’s challenge. This year, the questions focused on teachers and educators, workforce of the future, frontlines of health and coastal communities.
The competition is then opened up to participants from around the world with relatively low barriers to entry, resulting in 1,150 submissions from 110 countries in the last competition round. (That’s at least one submission from nearly 60 percent of all countries in the world.)
To qualify, though, participants need to have more than just an idea. They must have a prototype that works, be either in the growth, pilot or scale stage, and be tech-driven. Submissions are then evaluated by judges from across industry, intergovernmental organizations and academia to get to 15 finalists for each of the four challenge questions. These 60 finalists get a full day with judges to be asked in-depth questions and have their ideas evaluated.
The day after, with all the preparations completed, the finalists get three minutes apiece to present on stage. Crucially, instead of one winner, eight finalists are chosen for each of the challenge questions.
Each finalist receives an initial $10,000 prize, plus a pool of hundreds of thousands of dollars provided by partners including General Motors, the Patrick J. McGovern Foundation, Consensys, and RISE.
This year, for example, Ugandan health care startup Neopenda brought in an additional $30,000 in funding through Solve, from a UN program sponsored by Citi. An intelligent messaging app called TalkingPoints, meanwhile, received backing from General Motors and Save the Children to develop its personalized coaching technology for parents and educators. (You can see more details on this year’s winners and prizes here.)
As opposed to being a “one and done competition” where winning the prize money marks the end of the competition, managing director of community Hala Hanna tells me that the real work begins once the Solver teams are selected. Each qualifying Solver team gets 12 months of engagement and support from the organization. “Our value-add is providing a network, from MIT and beyond, and then brokering partnerships,” she explains.
Solve also produces a series of co-branded programs with other educational and nonprofit organizations around the world. As a result, the Australian government uses the platform to run a smaller-scale challenge focused on issues in APAC, while the Mohammed Bin Rashid Foundation is using it for a larger scale Global Maker Challenge.
Perhaps the biggest testament to the Solve method getting traction is its funders putting in even more cash in support. At the closing event on Sunday, an upbeat Matthew Minor, Solve’s director for international programs, took to the stage decked out in Solve-branded socks and a broad smile. He announced the winning finalists—and more funding opportunities. Two of Solve’s original backers, the Atlassian Foundation and the Australian government, are continuing to invest out of a standing $2.6 million budget for companies in the workforce track. RISE, a global impact investing fund, is putting an additional $1 million into companies focused on coastal communities.
The Australians have already put in funding to help past winners scale after the program. One of them is Ruangguru, a digital boot camp in Indonesia that gives youth dropouts resources they need to earn graduation certificates. The startup had reached nearly a million Indonesians prior to participating in Solve; through the program and the additional funding, it assisted more than 3 million Indonesian youth by the end of last year. Iman Usman, one of Ruangguru’s founders, tells me that Solve enabled them to enter into partnerships that helped them scale across Indonesia in a way they would have never been able to do on their own.
Solve has also been unequivocally good at ensuring diversity, both in its own staffing and—perhaps for related reasons—in those that are chosen as finalists. Of Solve’s 20 full-time staff, 14 are women, as are six out of the seven leadership team members and—by my count—at least seven nationalities from four continents are represented on staff.
The 33 Solver teams selected at the finals this year hail from 28 different countries, with 61 percent of them being women-led. At a time when the tech industry is struggling to increase diversity, Solve’s emphasis on diversity in challenge design and promotion has led to applicants and finalists that reflect the world Solve aims to help.
Hanna noted that increasing diversity is not as difficult as it’s made out to be. “Honestly, we’re not even trying that hard,” she explained. “So whoever says there are no women in tech, I say, crazy talk.”
Still, Solve does have a few kinks to work out. By taking on extremely broad topics, the competition can sometimes lack focus. Lofty questions mean you can get very disparate answers, making it hard to compare them in a way that feels fair. The work of the future challenge, for example, had one team pitching on adding jobs related to knitting in Brazil, while another managed to fit in every possible buzzword (artificial intelligence, crypto and automation) in three minutes without quite explaining what it does.
And while it’s great that the award monies are not all given to a single winner, it is not quite clear how funders pick the teams that do get funding. 15 qualifying finalists this year ended up winning money awards, some winning more than one, while the remaining 18 qualifying teams went home with the minimum amount. This is because Solve funders get to pick which of the teams that qualify at the finals get their respective monetary prizes. Of course, all 33 qualifying teams equally get to be a part of the Solve class with all the support and training that includes.
Another kink is the audience choice award—selected through open online voting prior to the finals—but not tied to any clear concrete benefit. Take the example of Science for Sharing (Sci4S), a Mexico-based startup that trains teachers to better engage students in STEM and has already reached nearly a million children across Latin America. It garnered 419 community votes in the Education Challenge, more votes than any other participant in the category, and handedly won the audience choice award, but ultimately was not selected as a Solver team. Another education startup, Kenya-based Moringa School, only got two votes but was selected. While Moringa and others were compelling and qualified in their own right, but it’s still hard not to think that Sci4S should have focused all of its time on its presentation and ignored the audience vote.
All in all, Solve does get a number of things right where other innovation challenges have failed. Instead of anointing one winner for the entire competition, it selects a class of dozens—reflecting the simple fact that the world’s most intractable problems are not going to be solved by any singular idea. Unlike many challenges put on by educational institutions and open only to their own students, Solve opens its doors wide. And winning at the finals doesn’t end your connection with MIT, it only starts it, with all qualifying finalists getting a year of individualized support, training and mentorship.
Done right, prizes can be effective at incentivizing startups to focus on pressing societal issues that can truly benefit from tech-drive solutions. But prizes for the sake of prizes can add to the noise and dissipate scarce public resources and entrepreneur attention. In the increasingly crowded world of innovation challenges promising to change the world, MIT’s Solve is a step away from the noise and towards effective prize granting.
Dressed in a Naruto t-shirt and a hat emblazoned with the phrase “lone wolf,” Ne-Yo slouches over in a chair inside a Holberton School classroom. The Grammy-winning recording artist is struggling to remember the name of “that actor,” the one who’s had a successful career in both the entertainment industry and tech investing.
“I learned about all the things he was doing and I thought it was great for him,” Ne-Yo told TechCrunch. “But I didn’t really know what my place in tech would be.”
It turns out “that actor” is Ashton Kutcher, widely known in Hollywood and beyond for his role in several blockbusters and the TV sitcom That ’70s Show, and respected in Silicon Valley for his investments via Sound Ventures and A-Grade in Uber, Airbnb, Spotify, Bird and several others.
Ne-Yo, for his part, is known for a string of R&B hits including So Sick, One in a Million and Because of You. His latest album, Good Man, came out in June.
Ne-Yo, like Kutcher, is interested in pursuing a side gig in investing but he doesn’t want to waste time chasing down the next big thing. His goal, he explained, is to use his wealth to encourage people like him to view software engineering and other technical careers as viable options.
“Little black kids growing up don’t say things like ‘I want to be a coder when I grow up,’ because it’s not real to them, they don’t see people that look like me doing it,” Ne-Yo said. “But tech is changing the world, like literally by the day, by the second, so I feel like it just makes the most sense to have it accessible to everyone.”
Last year, Ne-Yo finally made the leap into venture capital investing: his first deal, an investment in Holberton School, a two-year coding academy founded by Julien Barbier and Sylvain Kalache that trains full-stack engineers. The singer returned to San Francisco earlier this month for the grand opening of Holberton’s remodeled headquarters on Mission Street in the city’s SoMa neighborhood.
Holberton, a proposed alternative to a computer science degree, is free to students until they graduate and land a job, at which point they are asked to pay 17 percent of their salaries during their first three years in the workforce.
It has a different teaching philosophy than your average coding academy or four-year university. It relies on project-based and peer learning, i.e. students helping and teaching each other; there are no formal teachers or lecturers. The concept appears to be working. Holberton says their former students are now employed at Apple, NASA, LinkedIn, Facebook, Dropbox and Tesla.
Ne-Yo participated in Holberton’s $2.3 million round in February 2017 alongside Reach Capital and Insight Venture Partners, as well as Trinity Ventures, the VC firm that introduced Ne-Yo to the edtech startup. Holberton has since raised an additional $8 million from existing and new investors like daphni, Omidyar Network, Yahoo! co-founder Jerry Yang and Slideshare co-founder Jonathan Boutelle.
Holberton has used that capital to expand beyond the Bay Area. A school in New Haven, Conn., where the company hopes to reach students who can’t afford to live in tech’s hubs, is in development.
The startup’s emphasis on diversity is what attracted Ne-Yo to the project and why he signed on as a member of the board of trustees. More than half of Holberton’s students are people of color and 35 percent are women. Since Ne-Yo got involved, the number of African American applicants has doubled from roughly 5 percent to 11.5 percent.
“I didn’t really know what my place in tech would be.”
“When it was brought to my attention, I was like ‘ok, this is definitely a problem that needs to be addressed,’” he said. “It makes no sense that this thing that affects us all isn’t available to us all. If you don’t have the money or you don’t have the schooling, it’s not available to you, however, it’s affecting their lives the same way it’s affecting the rich guys’ lives.”
Holberton’s founders joked with TechCrunch that Ne-Yo has actually been more supportive and helpful in the last year than many of the venture capitalists who back Holberton. He’s very “hands-on,” they said. Despite the fact that he’s balancing a successful music career and doesn’t exactly have a lot of free time, he’s made sure to attend events at Holberton, like the recent grand opening, and will Skype with students occasionally.
“I wanted it to be grassroots and authentic.”
Ne-Yo was very careful to explain that he didn’t put money in Holberton for the good optics.
“This isn’t something I just wanted to put my name on,” he said. “I wanted to make sure [the founders] knew this was something I was going to be serious about and not just do the celebrity thing. I wanted it to be grassroots and authentic so we dropped whatever we were doing and came down, met these guys, hung out with the students and hung out at the school to see what it’s really about.”
What’s next for Ne-Yo? A career in venture capital, perhaps? He’s definitely interested and will definitely be making more investments soon, but a full pivot into VC is unlikely.
At the end of the day, Silicon Valley doesn’t need more people with fat wallets and a hankering for the billionaire lifestyle. What it needs are people who have the money and resources necessary to bolster the right businesses and who care enough to prioritize diversity and inclusivity over yet another payday.
“Not to toot the horn or brag, but I’m not missing any meals,” Ne-Yo said. “So, if I’m going to do it, let it mean something.”
Musk will resign from his role as chairman of the Tesla board within 45 days of the agreement, which was filed Saturday. He has agreed to not seek reelection or accept an appointment as chairman for three years. An independent chairman will be appointed, under the settlement agreement.
Tesla will pay a separate $20 million penalty, according to the SEC. The SEC said the charge and fine against Tesla is for failing to require disclosure controls and procedures relating to Musk’s tweets.
Musk doesn’t have to admit or deny the SEC’s allegations as part of the agreement.
Tesla has also agreed to appoint two new independent directors to its board and establish a new committee of independent directors and put in place additional controls and procedures to oversee Musk’s communications, according to the SEC. This likely means that Musk, who frequently turns to Twitter to unveils new products, features and updates on his multiple companies, will be more restricted moving forward. At least when it comes to his tweets about Tesla.
“The resolution is intended to prevent further market disruption and harm to Tesla’s shareholders,” Steven Peikin, co-director of the SEC’s Enforcement Division said in a statement.
The agreement marks the beginning of a new era of corporate governance for Tesla, which some shareholders have argued is too tightly controlled by Musk and others closely aligned to him such as his brother Kimbal Musk. Investor and founding board member Steve Jurvetson is still on leave.
In 2017, Tesla diversified its board and added James Rupert Murdoch, the CEO of Twenty-First Century Fox Inc., and Linda Johnson Rice,Chairman and CEO of Johnson Publishing Company.
Other board members include: Robyn Denholm, who joined the board in 2014, Brad W. Buss, who has been on since 2009, Antonio Gracias, and Ira Ehrenpreis, one of longest-serving board members who joined in 2007.
The SEC filed a complaint Thursday in federal district court alleged that Musk lied when he tweeted on August 7 that he had “funding secured” for a private takeover of the company at $420 per share. Federal securities regulators reportedly served Tesla with a subpoena just a week after the tweet. Investigations can take years before any action is taken, if at all. In this case, charges were filed just six weeks later.
The SEC said in the complaint that Musk violated anti-fraud provisions of the federal securities laws. The commission has asked the court to fine Musk and bar the billionaire entrepreneur from serving as an officer or director of a public company.
Musk described fraud charges an “unjustified action” that has left him “deeply saddened and disappointed.”
Tesla and the board later issued a joint statement supporting Musk.
The complaint contains a number of eye-browing raising details, including that he had talked to the board about an offer to take Tesla private as early as August 2 when he sent to Tesla’s board of directors, chief financial officer and general counsel an email with the subject, “Offer to Take Tesla Private at $420.”
Google’s big hardware event, scheduled for October 9, is expected to feature the new Pixel 3 and Pixel 3 XL phones. But now we know that Google will probably reveal a third-generation model of Chromecast, thanks to one recent Best Buy customer who discovered the device on store shelves.
“GroveStreetHomie” detailed his experience on a Reddit post entitled “I think I bought the 3rd gen Chromecast too early.”
According to the Reddit post, the customer went to Best Buy earlier to pick up a Chromecast for a new TV. That’s when “GroveStreetHomie” noticed the packaging and design was different from an earlier version.
The cashier wasn’t able to scan the item because it wasn’t in the system yet. The release date was labeled October 9 — the same day as the 2018 Google hardware event.
“But since I already had it in my hand and was the same price as the 2nd generation Chromecast, they let me have it under the old SKU,” the post read.
This new unannounced Chromecast is apparently thicker than the second-generation model. The Chrome logo has been replaced with Google one. The new device still has a micro-USB. The HDMI connector on the tip and base has been removed, according to the user.