American Airlines Kicked a Woman off a Flight and Called the Police. Here's Why the Airline Made a Mess of Things

Absurdly Driven looks at the world of business with a skeptical eye and a firmly rooted tongue in cheek. 

Never assume that your outbound journey on an airline will be subject to the same joys as the one coming back.

The crew will likely be different. Their moods will likely be different. And consistency in airline service isn’t so easy to find.

This is something DePaul University music student Jingjing Hu found to her cost.

On Thursday, she was flying back to her home in Chicago from Miami. 

She had perhaps her best friend with her — her cello. Hu had been performing in a music festival. 

Airlines allow musical instruments on board, as long as you buy them a seat and strap them in.

Musical instruments are like children, but infinitely more melodic and much quieter on planes.

As her husband Jay Tang described on Facebook, this was where the music became hard to listen to, as the mood, dynamics and tempo of the trip were derailed.

Hu was suddenly told that her cello was too big for a Boeing 737.


This was just before the plane’s doors were about to close.

Yes, she was already on the plane, having been through all the palaver of security. She was also allowed to pre-board and says she had been given a belt by cabin crew to strap the cello in securely.

Worse, as she was escorted off the plane, her cello allegedly brushed the pilot. He apparently felt hurt by this.

She took a picture of him as he displayed what some might take as a V for Victory sign, as Hu was ushered away. 

The airline insists he was signaling to ground staff that there were now two free seats on the plane. Airlines rarely miss a money-making opportunity, do they?

Could that have played a role here? The two seats were quickly filled.


This, though, was merely the first movement of many. 

Hu said she’d been told that she could catch the next flight, a mere hour or so later.

Alternatively, she could purchase a First Class seat. Well, two.

Ah, but that flight wouldn’t let her and her cello on. It was also a 737. And there, at the gate, there were three police officers. The airline had called law enforcement.

Then she says she was taken to a Holiday Inn — first the wrong one, then the right one — and only put on a flight the following day.

Even at this point, she says she was denied pre-boarding.


Of course, so far we only have her and her husband’s side of things. Could it be that she’d begun to show frustration at some point in this mess?

It would have been hard not to, perhaps. 

“You have so many chances to tell me that you cannot board yesterday,” she told NBC Chicago. “You never told me until I sat down.”

On Facebook, Tang offered dark motives for his wife’s expulsion: 

AA is just playing around with customers. They just kick off passengers when they oversell their tickets using FAA regulations as an excuse. I could have been told those regulations when purchasing the ticket. My wife could have been told those regulations when flying from Chicago to Miami, at check in counter in Miami International Airport, at the gate or even when boarding the plane.

American Airlines rules allow an instrument on a separate seat if it weighs less than 165 lbs. Hu’s cello weighs 10.

My own calculations suggest that the seats on an American 737 are actually more or less the same size as those on the 757 and 767 which also fly on the Chicago-Miami route.

So she should have been allowed on the second 737, never mind the first.


I contacted American to ask for its side of this troubling farrago. A spokeswoman told me: 

A passenger on flight 2457 from Miami to Chicago was traveling with her cello.  Unfortunately, there was a miscommunication about whether the cello she was traveling with met the requirements to fit onboard the particular aircraft she was flying, a Boeing 737. We rebooked our passenger on a flight the next morning on a larger aircraft, a Boeing 767.

A miscommunication.

And somewhere, the University of Euphemisms announced that its work was done.

The airline added that it had provided hotel and meals for the “inconvenience.”

My sources suggest the airline believes the problem may have begun with a customer service manager, not a Flight Attendant. This manager suddenly decided that the cello was inappropriate for the plane.

Or, rather, had decided the cello was actually a bass violin. 

The rules do say that cellos should be in a bulkhead window seat in an exit row.

Tang told me his wife was booked in 23A and 23B, not bulkhead seats. But the airline had told him and his wife in advance that there would be no problem. And the cello was strapped into the window seat, so it wouldn’t be in anyone’s way.

Tang said the airline told him it would perform a “deep dive” into the issue. He says it hasn’t contacted him since.

The customer service manager — if that’s the person at the heart of this — was clearly, though, a touch mistaken. There was nothing wrong with the size of the cello.

Some might wonder whether the pilot and/or the Flight Attendants would also have known the cello rules — or at least got the plane moving.

Instead, a scene.


When it comes to cellos, American has a discordant record.

Why, though, didn’t it dawn on the cabin crew and the pilot that Hu had been allowed onto the plane and, presumably, told there was no problem with her instrument? After all, she’d already been given a belt to strap it in.

Indeed, Tang says he’d made all the appropriate calls when booking to ensure that her cello was permissible.

At the core, then, is another example of airline personnel enforcing rules — worse, in this case, seemingly not even knowing precisely what they actually are.

Which can cause a little frustration for passengers. 

I’m sorry. I mean mishappiness.

Meet Serverless Inc. – The Startup That Aims To Accelerate Serverless Computing

Serverless computing is no more a buzzword. It has become a first-class citizen of the cloud alongside virtual machines and containers. Serverless Inc., a Bay Area-based startup realized the potential of serverless computing back in 2015 when the only prominent serverless platform was AWS Lambda.

Serverless PlatformSource: Serverless Inc.

Last week, Serverless Inc. announced that it had raised $10 million in Series A funding from Lightspeed Ventures. Including the seed round, the company has raised a total of $13 million.

Austen Collins, CEO and Founder of Serverless Inc., saw the opportunity in creating an open source tool to simplify application development targeting AWS Lambda. What started as an abstraction to Lambda transformed into a cloud-agnostic serverless platform supporting all major serverless environments.

Serverless developer community loves the framework and tools shipped by Austen’s team. Apart from getting featured on the home page of Hacker News several times, this open source project on Github gathered about 25000 stars which a remarkable feat for an independent open source project.

Within a short span of time, Austen’s framework has become an unofficial standard for serverless tooling. Every major serverless platform debuts with support for Serverless Framework. From Google Cloud Functions and more recent Kubeless, almost all the major environments are supported by this framework.

So, what excites developers about Serverless Framework? It’s the simplified approach to building modern applications based on integrated tooling.

Unlike other application development and deployment patterns, serverless platforms are highly distributed. For building a simple web or mobile app, developers will have to assemble a variety of resources ranging from an API Gateway to object storage to an event-driven messaging service. They have to connect the dots across a variety of cloud services making deployment and debugging very complicated. Serverless Framework is one of the first toolkits that aims to simplify the lifecycle of microservices targeting serverless environments.

Recently, Serverless Inc. announced an end-to-end platform that comprises of the framework, a visual dashboard and a cloud-agnostic event gateway. At the core of the platform is the popular Serverless Framework that provides a declarative mechanism to deploy and manage functions. The Dashboard enables developers and operators to monitor, collaborate and secure serverless applications. The Event Gateway offers integration infrastructure by connecting serverless applications and functions to existing workloads and cloud services.

Serverless Platform attempts two things – plugging the gaps existing within mainstream serverless environments and delivering a consistent workflow. This approach minimizes the risk of adopting serverless environments for enterprises. I feel the uniqueness of Serverless Platform lies in its platform-independent approach. In its current form, it supports AWS, Azure, IBM Cloud, GCP and Kubeless among other environments. Developers follow the same workflow to manage the application lifecycle irrespective of the deployment target.

Serverless Inc., boasts of impressive clientele. EA Sports, Coca-Cola, Nordstrom, Expedia and Reuters are some of the big brands using the framework. The company has set an example for other startups aspiring to build an open source platform that enjoys the attention of the developer community while being highly relevant to enterprise customers.

Serverless computing and Functions as a Service (FaaS) are in their infancy. The market is extremely fragmented with dozens of services and tools. The industry has no standard for eventing, messaging, function declaration, versioning, deployment, monitoring and logging.

The container world is equally excited about the convergence of Kubernetes and FaaS. Multiple projects claim to deliver serverless capabilities on Kubernetes.

CNCF, the custodian of Kubernetes and related projects, is working on standardizing some of the building blocks of the serverless platform. The serverless working group at CNCF recently proposed a standard called CloudEvents for exchanging messages across functions. But, it has a long way to go before the vendors accept it as a standard.

With the fragmentation and complexity revolving around serverless computing, companies such as Serverless Inc., have a massive opportunity in building tools that aim to make developers productive in building modern applications.

All The Reasons Tesla Stock Popped After The Q2 Conference Call


The following is an analysis of the Q2 conference call. Specifically, bullish takeaways that I noted while listening to the call. The 16% gain on the day was due to multiple factors, which I will cover in some detail in this article.

Increasing Production rates

Musk reported that Tesla produced 5K/wk multiple times in July, which means it was not a sustained average but only burst rates, most likely extrapolated from individual daily rates.

He estimates a 5K/wk average for Q3, in line with the expected 6K/wk burst rate, which would make up the ground lost in July. The estimates of 50-55K for the quarter don’t add up to a 5K/wk average but may be factoring in some downtime for improvements.

The current bottleneck keeping the company from a 5K/wk steady state is body production, which it estimates will be solved in 1-2 weeks. Tesla says it achieved a burst rate of 800 bodies made in 24 hours. The production rates are clearly making significant progress on a quarterly basis, and any deadline misses from internal guidelines are secondary to actual output and deliveries which will improve Tesla’s bottom line for Q3

Surprising Sales and Demand

Model 3 US market share for the midsize luxury category surpassed all competitors combined. This was outlined in the shareholder newsletter but also mentioned again in the call. This means the majority of cars sold in this category were Model 3s.

Tesla’s head of sales says Model 3 trade-ins are also coming from non-mid-sized luxury owners. This means the addressable market for Model 3 is much larger than most people anticipated and not limited to luxury car buyers only. The company listed the top 5 vehicles that Model 3 buyers traded in between January and July: they were the Toyota Prius (NYSE:TM), BMW 3 Series (OTCPK:BMWYY), Honda Accord (NYSE:HMC), Honda Civic, and Nissan Leaf (OTCPK:NSANY). Tesla was especially surprised, since all but the BMW 3 Series are budget increases. Customers were trading up into a Model 3.

Musk noted that with an absence of a traditional sales team at showrooms, customers are the primary sales force. Viral sales are occurring by word of mouth. The more cars the it delivers, the more owners show off their cars, the more cars company sells.

It was noted that not all stores have test drives available yet. Tesla believes this lack of ability to test-drive the Model 3 has throttled potential orders. I’ve written in previous articles that this will be a significant lever that can be pulled to increase demand. Three weeks ago, eight stores had demo vehicles for test drives, and there are now ninety stores offering test drives.

As was predicted by Ben Sullins’s Teslanomics configurator, the company reports >50% of customers are choosing AWD and Performance versions over the RWD, leading to higher margins and ASP. This also dispels the narrative that the majority of buyers are waiting for the SR RWD base model, when the data is indicating more than half will opt for the AWD version of the SR battery.

Tesla also reported that when reservations were lifted in North America, new orders outnumbered converted reservations, indicating there is strong demand outside of the reservation list, likely due to the appeal of a shorter wait time. I noted in a previous article that once wait times were reduced, stronger demand would surface.

In July alone, 60K test drive requests were made. Tesla notes that the conversion ratio on test drives is strong. I also predicted in a previous article that test drives would add a considerable layer of demand given that 20% of all buyers ranked test drives among the top 3 reasons they bought their vehicle.

Positive Cash Flow and Sustainable Profitability Outlook

Musk estimates that Q3 and on, Tesla can achieve sustainable profitability and positive FCF. Recession or force majeure to the supply chain are only threats to positive FCF and GAAP profitability from here on. Quarters with loan paybacks could be an exception.

The potential production output of the lines is much greater than Tesla expected, and with fine-tuning of the production line from manual to automated and vice versa, with very little capex, margins have increased.

In addition to gross margin improvement, the company forecasts S/X volumes increasing in H2 as contributing to positive cash flows.

CFO Deepak Ahuja mentioned that opex is essentially flat with increasing revenues from the other businesses.

In response to analysts’ questions Tesla reiterated that it has no plans to raise equity. The plan is to use local debt to finance GF3 instead of tapping capital markets. The company clarified that all future plans are self-funded, despite plans for debt repayment to reduce interest expense.

Tesla says it has no cash flow issues, and that Q3 GAAP profitability is still on track as of the end of July. The company also noted that profitability will not come at the expense growth but expects certain quarters to be just above breakeven during factory building or loan repayment quarters.

Improved Autopilot With FSD Features

Tesla’s Autopilot (AP) team reports radical advances in neural net development and FSD chip development, with an order of magnitude improvement in operations per second. The current Nvidia (NASDAQ:NVDA) GPU the company is using is capable of 200fps, versus Tesla’s proprietary chip which is capable of 2000fps with redundancy, at the same cost. This new V3 hardware will offer full frame rate, full resolution, and full NN processing with idle cycles to spare. It is also plug and play for the existing vehicle interface.

(Source: Daily Mail)

Autopilot V9 is the release that was mentioned for some time in August. Features include navigation integration with autonomous lane changes, GPS routes, and on-ramp and off-ramp capabilities. It will also add new safety features due to nuanced perception of the vision system.

One surprise was that Tesla revealed this system has been running in stealth mode for the last 2-3 years, and the company believes it is finally safety-validated for deployment. FSD features are being deployed in iterations of AP as safety features first. The company is forecasting coast-to-coast potentially by EOY, but AP V9 release is the focus. It is estimating this breakthrough release in about a month.

Musk’s Improved Behavior

Musk apologized to Bernstein analyst Toni Sacconaghi for his treatment of the analyst on last quarter’s conference call. He mentioned a lack of sleep and being overworked at the time of that call. Musk also apologized to RBC analyst Joseph Spak for similar treatment.

(Source: Spaceflight Insider)

This is not a trivial detail, as most analysts were predicting a crisis of confidence if Musk repeated his behavior from the 1st quarter call. This more professional attitude is a comforting sign to shareholders that the company is still in good hands and the risk of Musk being removed from the helm is significantly reduced.

Positive Gross Margins On Model 3

Gross margins (GM) for Model 3 turned positive in Q2. Musk explains that Model 3 GM increases were due to the elimination of inefficiencies and bottlenecks. Labor cost per vehicle is improving, along with manufacturing cost efficiencies. CFO Deepak Ahuja also noted that initial ramp-up costs were one-time affairs, and that fixed costs are leveraged over higher volumes – which is to be expected. Musk noted that big expenses included low-volume tooling for emergency fixes. GM will continue to improve as efficiency improves without further capex.

In the Q&A, Tesla confirmed that 25% Model 3 GM would be at the lower $44K ASP by about Q2.

Factory Expansion

Tesla’s first Gigafactory, GF1 which was estimated at $5 billion, was a huge learning lesson for the company in terms of cost efficiencies, and the planned factory for Shanghai, GF3, would be more than 50% cheaper at around $2 billion. This would add capacity of about 250K vehicles/year, including battery module and pack production. This cost does not include cell production, as it would most likely be covered by another partnership with Panasonic (OTCPK:PCRFY).

(Source: Reuters)

Tesla estimates that it will be able to announce the location of GF4 Europe before EOY. The company reiterated that no capital raise was necessary, and that it would most likely use local banks for financing.

Massive Production Efficiencies

CTO J.B. Straubel noted that certain areas of production that were capital-intensive saw a 25-30% increase in output with zero additional capex.

Musk clarified that the Sprung structure is often being used as a permanent structure and is not the temporary “camping” tent it is being made out to be by critics. GA4 is permanent for now, but Tesla likes that it is flexible and iterative so it can be expanded or reconfigured. It was a capital-efficient solution that fit the company’s needs at the time. Musk noted that steel frame buildings are still the goal when asked if this would be a model for future Fremont expansion. The company did admit that the production line inside the tent would serve as a model for GF3, particularly the parts delivery workflow.

Tesla noted that GA4 re-used the conveyance system from previous S/X/3 production lines, which kept costs for GA4 down. One of its creative solutions included implementing a 1% grade so that it was downhill trip for the cars, in order to keep the line within the specified load rating of the conveyor.

(Source: @elonmusk)

The quality assurance team installed at the end of line was able to produce the same quality from GA4 as was achieved on GA3. All P3D vehicles are being made on GA4, which are the most expensive Model 3 vehicles you can buy. This is especially significant as it reduces rework costs as well as warranty service costs.

The simplicity of the GA4’s layout means fewer labor hours per car than GA3. This is due to a simpler parts delivery system, which allows trucks to drive up to the exact station that requires the parts. Currently, the parts for GA3 are unloaded into the warehouse and reloaded into bins for conveyance to the relevant station. This means that parts are unpacked, packed, and then unpacked again before being installed on the car.

Some non-essential GA3 robotics that required additional maintenance were removed, thus reducing maintenance costs for the line.

Both Musk and JB stressed that GA4 is not a walkback from automation, but selective automation would be applied, further reducing labour hours per vehicle – a metric often cited by shorts.


In response to an analyst’s question, Tesla clarified that it treats manufacturing problems as technology problems. Musk added that most production problems are software-related at volume.

The company blamed outsourced software development as the cause for the original issues in the battery pack assembly. These outsourced suppliers were supposed to be world-class experts but failed. Tesla rewrote the software to be more efficient and simpler.

Rapid iteration between design and production was key to unlocking bottlenecks. Tesla noted that its strategy is understanding the rate limiters of existing software. This allows the company to produce batteries that are lighter, better, cheaper, and with more range. It is estimating this to be active in about 6 months. The design engineers are working with automation and line engineers to improve production, which is allowing for better design for production.

The ability to problem-solve manufacturing as a technology problem keeps Tesla within its circle of competency and will lead to continuous production efficiencies and cost reductions.

Product Pipeline

In response to a question, Tesla said it has trimmed 2020 vehicle production estimates to between 700K and 800K. This includes 600K between Fremont and GF1 and 200K from Shanghai. It seems that being more realistic is looked on favorably by shareholders.

Plans for Model Y, Semi, Pickup, and Roadster were confirmed, but the company could not announce production locations.

When asked about Powerwall and Powerpack being such a small proportion of GF1 output, Tesla responded that the growth restrictor on those products is cell production. Panasonic recently just announced a 30% bump in output, which should help with this. Tesla Energy also relies on other cell suppliers like Samsung (OTC:SSNLF) and LG (OTC:LGEAF), which are supply-constrained now.

Tesla’s additional cell capacity at GF1 will allow it to ramp energy storage 3x-4x in 2019.

(Source: Teslarati)

The company is also ramping GF2 on solar and solar roof production. Tesla says there are several hundred homes with solar roofs. But it is still validating the longevity of the product, as well as ensuring first-responder training is enough to handle the incoming installations. This is probably why critics feel like solar roof is still vaporware.

The Energy division is constrained by a lack of certified electricians to install the systems. There aren’t enough certified electricians to handle the current delivery load.

Tesla confirmed that Semi will be using Model 3 components, such as motors, handles, and screens, but declined to give away any more details other than more improvements have been made since the reveal back in the fall. This repurposing of Model 3 components reduces R&D costs on Semi and improves economies of scale on the lower-volume truck prototype.

The company reaffirmed that it expects energy and auto revenue will reach parity.


There was immense concern by critics that Tesla was going to need a capital raise to fund future projects, and also regarding Tesla’s guidance for consistent quarterly profitability, and thus, annual profitability put this to rest. Musk’s behavior was also a concern for critics, but the call took place in a civilized and professional manner. Gross margin improvements due to improving production efficiencies are paving the way for the full Tesla product pipeline to become a reality.

The stock finished the day up 16.9%.

Disclosure: I am/we are long TSLA.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Biotech Analysis Central Pharma News: Teva's U.S. Erosion, Denali's Preliminary Results, Epizyme's Trial Shutdown

Welcome to Biotech Analysis Central Daily News, a daily news report and analysis about what has happened lately in the biotech industry.

Teva Dips As U.S. Generic Sales Continue To Drop

News: Recently, Teva Pharmaceuticals (TEVA) announced its Q2 earnings. It reported adjusted income of $0.78 cents per share on sales of $4.7 billion. In comparison analysts were expecting the company to earn $0.67 cents per share on sales of $4.76 billion. Generic U.S. Sales dipped down to $947 million for the quarter. It was also reported that sales of Copaxone dipped by 46% to $464 million. It was revealed that debt remains at around $30.2 billion.

Analysis: The earnings per share number was a beat, however sales came in slightly below the consensus of what was expected. However, setting that aside there are major issues that remain. The first issue is that earnings have still been declining year over year. In the same period last year, Teva reported adjusted earnings of $1.02 per share on sales of $5.72 billion. That’s a huge dip year over year in both EPS and in revenue from the numbers reported. Generic sales in the U.S. continue to erode due to increased competition from copycat drugs. It is just the landscape of the generics market which is bad and continues to fall. It is even said that generic U.S. drug sales fall an average of 15% per quarter. In my opinion, such a drop is not sustainable. Copaxone continues to fall by the wayside as a generic versions of the drug continue to eat into sales. Teva attempted to counter the eroding sales with laquinimod, but that has failed to live up to expectations in Multiple Sclerosis (MS) and in other diseases. Just recently, Teva and its partner Active Biotech (ACTI) failed a phase 2 study using laquinimod to treat Huntington’s disease. Debt is slowly coming down, but in my eyes it’s not fast enough. That debt of $30.2 billion will still continue to be a major risk factor and burden on the stock.

Denali Therapeutics Shows Positive Preliminary Data For Parkinson’s

News: Recently, Denali Therapeutics (DNLI) announced that it had obtained positive results from its phase 1 study treating patients with Parkinson’s using DNL201. This phase 1 study recruited more than 100 healthy patients who received either a single or multiple ascending doses of the drug. In addition, some patients in the study were given placebo. It is believed that DNL201 is able to inhibit leucine-rich repeat kinase 2 ((LRRK2)). The reason for doing so is because it is believed LRRK2 regulates lysosomal genesis and function, and for those with Parkinson’s this mutation causes the disease. The drug was shown to encourage robust target engagement in two blood-based biomarkers of LRRK2 activity. DNL201 was also shown to have an effect on biomarkers for lysosomal function. These may have some indication that this drug can work in this population.

Analysis: These are preliminary findings, but the most important item to note is that targeting LRRK2 activity may be a way to treat Parkinson’s. There is another item to make note of, and that is the ability for the drug to be adaptive. By that I mean even though DNL201 targets patients with the LRRK2 protein, it can also be used in a broad sense. That means DNL201 is going to be explored in a phase 1b study as the next part of its clinical advancement. It will treat Parkinson’s patients with or without the LRRK2 protein. This study is expected to start by the end of 2018. This is a good start on preliminary analysis, and the next study should incorporate a higher dose that should improve efficacy.

Epizyme Shutters B-Cell Cancer Study After Trial Observation

News: Recently, Epizyme (EPZM) released its Q2 earnings report and updates on its pipeline. It was noted in the report that it would stop development of its drug tazemetostat for treating patients with diffuse large B-cell lymphoma (DLBCL). It was noted that in cohorts from a particular study, tazemetostat was not performing well in treating patients with DLBCL. This observation was made with both tazemetostat as a monotherapy and as a combination with prednisolone.

Analysis: This is another huge blow for Epizyme, because tazemetostat is the main drug in its pipeline. This bad news is in addition to another prior issue where tazemetostat was placed on an FDA partial clinical hold for a study treating patients with genetically defined solid tumors and malignancies. The reason for the partial clinical hold was that a pediatric patient developed secondary T-cell lymphoblastic lymphoma. This prompted the FDA to immediately place the partial clinical hold on the trial. It has been about 3 months now since the hold was announced, but I guess the bright side is that it was only a partial hold. Meaning, patients that were already in the study could still receive treatment with tazemetostat. In essence, the hold was just to stop the recruitment of new patients. Even setting the issue with the FDA clinical hold aside, Epizyme having to dump its DLBCL indication due to lack of efficacy is not good at all. The hope now lies in the ability for Epizyme to get the partial clinical hold lifted by the FDA. If it can do that in a timely manner, then maybe it still has a shot to turn things around.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Google, Facebook Aim To Fix AI Diversity Issue By Funding African Machine Intelligence Course

SHANGHAI, CHINA – JULY 04: An AI robot by CloudMinds is on display on the opening day of the China International Robot Show 2018 (CIROS 2018) on July 4, 2018 in Shanghai, China.  (Photo by Tang Yanjun/China News Service/VCG)

Google and Facebook are funding a machine intelligence masters course in Africa as part of an effort to fix the industry’s diversity problem.

The African Institute for Mathematical Sciences (AIMS) will launch the “African Masters in Machine Intelligence (AMMI)” course in Rwanda in September, before looking to roll it out to other African nations. 

AMMI will be a one year course that will train young Africans in machine learning and its applications, according to a blog post on Medium by Moustapha Cisse, who is the founder and director of the course. Cisse added that Facebook and Google’s sponsorship and partnership are making AMMI possible.

Facebook said it will provide $4 million in funding and staff lecturers over several years, while Google is also contributing resources.

Our investment and engagement in AIMS is part of our commitment to the creation of a more diverse talent pipeline,” Facebook Research wrote in a blog post on July 31. “It is notable that the talent pool currently actively researching this space has remained relatively small and, importantly, is unrepresentative of the diversity of our world in terms of ethnic representation as well as geographic, economic and cultural background.” 

Jeff Dean, head of Google AI, wrote on Twitter: “I’m very excited to have Google and Facebook come together to help fund this new Masters program at the African Institute of Mathematical Sciences (AIMS).”

Google did not immediately respond to a request for comment.

I cover artificial intelligence and Google DeepMind. In previous roles I’ve written about tech policy, European startups, the gig economy, and venture capital. Today I’m a freelance journalist and I’m also working on a book about Google DeepMind. Previously I was a senior te…


'Madden 19' Review: The Good, The Bad And The Bottom Line

With an earlier release date than normal, Madden season has begun a little sooner for fans of the historic football gaming franchise.

Lamar JacksonCredit: EA Sports

The leadership situation on the development team is in a transitional state. Thus, there are a few legitimate concerns about this year’s finished product and the series moving forward. The game goes live for Xbox One EA Access subscribers late on Wednesday, and there are already a number of fans getting a taste of this year’s release.

What should you expect from this year’s game? There are some areas that are the equivalent to a fly route that turns into touchdowns. Others are akin to infuriating five-yard penalties that stop a drive cold. If you’ve played Madden for years, you know what I mean.

The GoodCredit: Brian Mazique

Visuals…The Good Parts

Dak PrescottCredit: EA

I’ve thought about this next sentence for the past 4 days because I didn’t want to sound like a prisoner of the moment. After further review, I feel confident in saying: Madden NFL 19 is the best-looking sports video game I’ve ever played.

Let me put that into proper perspective. I’m talking purely about the visuals as it pertains to the overall look of the players, equipment, stadiums, grass, lighting, etc. I’m telling you, the players and objects in this game are rendered so nicely, it’s hard not to recognize the quality. You can take almost any screenshot, zoomed in or out, and marvel at the level of detail in almost every aspect of this virtual world.

Quentin NelsonCredit: EA

I’m more than 15 games in and I’m still taking the time to stare at the reflection in the helmets. Even the cutscenes, overlays (though there could be more from a statistical standpoint) and monoliths look good. The art team and whoever is responsible for the visual quality of this year’s version of Madden deserves a round of applause because those men and women did an amazing job.

This is easily the best example of what the Frostbite Engine has done for an EA Sports title.

Animations…The Good Parts

There are some nice new animations in the game. The dropped passes look a lot better than they did before. Also, the tackles are impacting, especially the ones involving three or more players.

This version of Madden captures a nice balance between hard and fun-to-watch hits and the safer, softer tackles that we saw in recent versions of the game.

Some of the interception and fumble recovery animations are incredibly smooth. I had one memorable fumble recovery on the sidelines. A linebacker in zone coverage put a solid hit on a receiver just as he was trying to run out of bounds. The ball popped in the air, and my safety ran over, caught it in the air with a seamless animation and headed the other way with the ball.

It was a satisfying play, not only because my defense created a takeaway, but more importantly, I could see exactly what happened during the instant replay. That’s the way it should be in a sports video game. Quite a few of the virtual athletes have their moments in the game, but running backs look the smoothest in action in Madden 19.

LeSean McCoyCredit: EA

The jump cuts, hesitations, and spins will likely cause the most oohs and ahhs from fans.

The Positive Effects of Real-Player Motion

RPM can be felt in Madden a little more than in FIFA 18. Quite honestly, I wasn’t a big believer in the technology until EA UFC 3 was released, and then I started to pay attention to what the possibilities could be with the feature.

It’s not something that should be described as a gamechanger in Madden 19, but you can feel the weight and speed of players more definitively. That’s a good foundation.

Stick to the Numbers

The ratings seem to matter more in this year’s game than in other versions. You can feel the difference between elite players and average to below-average guys, and that’s not just at the skill positions. Offensive lineman and pass rushers who aren’t especially good will pale in comparison to the top guys.

Carson WentzCredit: EA

When building a team in CFM and MUT Draft, you’ll need to be aware of your impact players and the positions they play.

MUT is Strong

Ultimate Team continues to be one of the strongest–if not the strongest area of the game this year. Solo Battles brings a favorable offline element to the mode. Players will pit their teams against random lineups plucked from user’s collections with varying difficulties and rewards. While the gameplay is technically offline, Solo Battles still has an online component. There are 14 tiers of rewards in the mode with 13 total games each week.

Solo BattlesCredit: EA

The objective is to win the game and score as many points as possible. It’s a fun new wrinkle to the MUT system. It may need some further development, as was the case with MUT Squads last year. This year, MUT Squads against the A.I. and MUT player upgrades round out the newest additions to Ultimate Team.

In MUT Squads against the AI, you and friends have an opportunity to play for rewards and upgrades against CPU opponents in challenges.

Aside from providing a way to earn unique content, this new angle on co-op play gives you and your teammates a chance to practice against an opponent. Co-op Madden is more challenging than just about any other sports video game, but when things come together, it can be fun.

Create-a-Draft Class

Draft ClassesCredit: EA

EA has finally given fans the ability to create their own draft classes for Connected Franchise Mode. This is perhaps one of the most overdue additions on the current sports gaming landscape. I’m happy to confirm that the feature is as solid and in-depth as was originally advertised.

This should help to add longevity to the overall CFM experience.

Player Development and Scheme Fit in CFM

Scheme FitCredit: EA

This aspect of the game might go unnoticed by fans who won’t take the time to recognize the substance and what it can do for CFM. The entire player development, scheme fit system works really well in Madden 19. Not every player develops or regresses at the same pace, and the game does a good job accounting for those factors. Training players seems a little more relevant this year too.

Player performance can be impacted by how well a guy fits into his coach’s scheme. If you take the time to learn the kinds of players you need to make your system successful and then target players who fit the bill in free agency. trades and the draft, you’ll see how it all comes together.

This was a good addition to the game’s feature set.

Longshot…The Good Parts

Longshot 2Credit: EA

Devin Wade and Colt Cruise are back for Longshot 2: Homecoming. The voice acting is really strong again this year, and Rob Schneider’s character is hilarious. More to come on this mode a little later on, though.

The BadCredit: Brian Mazique

Animations…The Bad Parts

In some ways, Madden 19 is a victim of its own beauty. The game looks so good in still shots, slow motion, and most live action that when you see poor animations, they look even worse because it’s happening with such accurately rendered models.

Some of the things I saw during my review period might be eliminated with a Day 1 patch, but I have to report what I saw in my time with the game. There were some incidents of frame skipping, and some really strange things occurring after injuries.

On one play, Ty Montgomery got injured and somehow he got stuck in the hurdle animation. He hilariously hurdled his way off the field. He didn’t come back, but I’m pretty sure that wasn’t the way he was supposed to exit the game.

There’s also a pretty significant issue on punt returns. The CPU returner will sometimes touch the ball and leave it for your coverage team to recover. That led to a TD and possession deep in the CPU’s territory on separate occasions.

Again, this is another area of the game that could be patched soon.

Real-Player Motion…The Bad Parts

As I mentioned earlier, RPM can be felt in the game, only it’s not always good. It takes some getting used to as the locomotion of the players–especially ball carriers–can be herky-jerky. Sometimes it feels as if the movement system creates an extra step or motion that seemed to cost me in a few instances as I tried to hit a hole.

This is something that I might get a better handle on after more reps, but chances are I’ll still feel as though it could be tuned.

Low Situational A.I. 

Sideline player A.I. is still an issue for what seems like the 15th consecutive version. On one play, I called an out route for Larry Fitzgerald. I threw a bullet pass to him immediately out of his break, and he just kept running clean out of bounds and off the screen entirely. The next day it seemed this issue had been resolved, but receivers still aren’t respecting the sidelines or making a reasonable effort to stay in bounds.

It’s pretty frustrating we’re still seeing this issue after so many years.

The CPU’s playcalling at the end of the half and game is worst than Darrell Bevell’s in Super Bowl 49. The CPU will throw the ball in obvious run situations, and I’m not talking about taking reasonable gambles; I’m talking decisions that just go against the fabric of the NFL and good football.

CFM Still Needs Some Work

The addition of created draft classes, player development and offensive and defensive schemes is nice, but CFM is still a ways behind NBA 2K‘s MyLeague and MyGM, and even MLB The Show‘s franchise mode. For starters, the injury presentation is still lacking.

We’re still getting a full diagnosis of injuries within the game the injury took place. That’s not at all realistic. Can we get a virtual MRI that at least forces you to wait until the end of your game to find out the details?

Also, the off-ball injuries are still phantom events with no visual representation. Because of the way they take place, they are more irritating than immersive. These are essentially randomized injuries that the system alerts you to during games. If you go back and watch the replay, you’ll see, the player shows no sign of injury at all.

Injuries also seemed to happen at a relatively high rate in at least of one of my CFMs. I started one with the Chicago Bears and I lost two players for the season within the first few weeks of the regular season, and at least one player for multiple games in every contest.

I was happy to see that didn’t happen as much in the first 3 weeks of my second CFM. I used a fantasy draft for that one because I wanted to see what the game felt like playing with better players than my Bears have at their disposal. I’m still not sure if the injury rate is too high, at least on default settings, but this is something you can tweak in the sliders. However, I don’t think you should have to do that. The injury settings should be on the most realistic level as a default.

Stat tracking still has some problems, and this may be one of the more egregious issues. Mostly because it’s been a problem of sorts for a few years now. Some sacks still aren’t recorded or even called out correctly by the commentators. The first time it happened, I thought I was mistaken. Thankfully, I was capturing the footage and confirmed on multiple that not only did a player not get credit for the sack a few times, play-by-play commentator Brandon Gaudin called out the wrong guy in his dialog.

Presentation Problems

The commentary is still OK, but this is the first year I’ve started to feel as though Gaudin and Charles Davis’ lines are getting redundant. It’s not the basic lines, it’s the longer spiels that sometimes don’t quite match the sequence on the screen. I think it’s important to keep the longer stories fresh because those are the ones that are the most memorable. The moment Davis says, “I used to have a teammate who used to say, I hate experienced quarterbacks,” I’m turned off because I heard this same speech a million times in Madden 18.

It’s not ideal to duplicate such a recognizable soliloquy.

I suppose there is something to be said for effort, but the new halftime show feels undercooked. It’s as if they wanted to show highlights from other games, but ran into some issues and decided to leave the bland stats and scores on the screen with the cool 3-D map. Jonathan Coachman replaces Larry Ridley as the voice of the halftime show, but it doesn’t really impact the presentation’s effectiveness.

Longshot Gets a Little Long and Cheesy

While the story is well written again this year, and voice acting is still strong, things get a little cheesy about ¾ of the way through. I won’t give any spoilers, but if you play through the entire mode, you might see what I mean. It starts to feel like an after-school special from the 80s or 90s.

This section of the story could have been made about 20-30 minutes shorter, and that would have added a little more value. Sometimes less is more.

The Bottom LineCredit: Brian Mazique

This year, Madden doesn’t break any new ground or fix all or most of the nagging issues that have impeded the series in the past. However, it does deliver an absolutely stunning visual experience, decent gameplay and enough options to keep most fans occupied.


Developer: EA Sports

Publisher: EA Sports

Platforms: PS4, Xbox One (version tested), PC

Release Date: August 10, 2018 (EA Access begins on Aug. 2 and early access for pre-orders starts on Aug. 7)

Price: $59.99

Score: 7.7/10

Disclaimer: Free review code was provided for coverage purposes

Here's Who Wins When Blockchain Meets the Food Chain

Margins are razor-thin in the food industry. So when a bacteria outbreak like E. coli hits, it can really wreak havoc throughout the entire supply chain.

Now, 10 leading food companies plan to build a digital tracking system that’s “the equivalent of FedEx tracking for food.”

That’s how Frank Yiannas, vice president of food safety at Walmart, described it to The Wall Street Journal.

Senior Vice President, IBM Global Industries, Platforms and Blockchain Bridget van Kralingen speaks during the forum Digitalization and the New Gilded Age at the World Bank/IMF spring meetings on Wednesday, April 18, 2018, in Washington. ( AP Photo/Jose Luis Magana)

He’s being coy. It’s a revolution. And investors should take note.

Meet the New ‘Food Trust’

Under Walmart’s leadership, Nestlé SA, Dole Food Co., Driscoll’s, Golden State Foods, Kroger Co., McCormick,  McLane Co., Tyson Foods,  and Unilever, are attempting to do something that has never been done before.

These food industry giants want to create a food trust. They want track food from the farm to the grocery aisle.

If something goes wrong, they will know exactly where it began, and why.

That type of transparency would forever change the food supply chain. It would keep suppliers on their toes.

One wrong move could kill years of goodwill. It would also give distributors and processors unprecedented control.

o make this possible, the food companies will use blockchain. The distributed digital ledger, best known for its role in keeping cryptocurrencies in check, will do the same for the food supply chain.

It will create an electronic verification network in real-time for every single food product in the trust.

Blockchain: The Missing Link in the Food Chain

Blockchain is uniquely suited for this task. That’s because its ledger system is permanent and can’t be altered.

Every time a verified event occurs, a block is created on the chain. And it is viewable by all. That makes the system inherently “trustless,” which in this context means it does not require trust.

Blockchains are popping up everywhere …

In 2017, the Financial Times reported six of the world’s largest banks — Barclays, Credit Suisse, CIBC, HSBC, MUFG and State Street — announced support for the Utility Settlement Coin (USC). That’s a blockchain created by UBS, the Swiss banking giant.

This USC system is supposed to let banks conduct transactions without waiting for traditional money transfers.

In January 2018, Maersk, the container ship conglomerate, announced a blockchain for transoceanic logistics. The Danish company hopes this open-source, digital platform will become the standard for a $4 trillion shipping industry that’s currently drowning in bureaucratic red tape and piracy.

International Business Machines is winning the public relations war. It’s behind the Maersk and food trust blockchains.

However, investors should look elsewhere for blockchain winners …

My research suggests they should focus on Microsoft.

Unlike IBM, the Redmond, Wash., software giant has the cloud computing scale to be a dominant player in blockchain. It also has a history of solving big, real world problems with the technology.

In 2015, the company shifted its business model from selling software licenses to digital subscriptions. The distinction created a flood of partners looking to sell cloud services on Azure, its cloud network.

Microsoft needed a system to quickly assess creditworthiness. The traditional process involved working with individual banks to issue standby letters of credit for the new sellers.

But the process was antiquated, and involved several levels of manual verification. So Microsoft worked with Bank of America to digitize and automate the entire process with blockchain.

Ethereum Enters the Picture

Since 2015, the company has expanded its work with blockchain …

It partnered with ConsenSys, a New York blockchain software company, to bring the Ethereum blockchain to Azure, as a Software-as-a Service layer.

Microsoft leveraged this modularity to win over R3, a consortium of 200 financial institutions. Together, they are developing Corda, a blockchain specifically built to reduce transaction friction in financial services.

In 2017, Bain & Co., a research and analytics firm, estimated $35 billion in operating and capital cost savings for a blockchain of this scale.

For Microsoft, Corda is the gateway. The prize is untold billions in sales of cloud computing services.

It is a big business for Microsoft. In the first quarter of fiscal 2018, Azure commercial cloud revenue jumped 58% to $6 billion.

As businesses move to the cloud, having a blockchain software module is a huge advantage.

The prospects are being reflected in the price. Since 2015, shares have risen an average of 31% annually. And the real growth lies ahead as more companies embrace the cloud and SaaS applications.

FedEx-like tracking for everything is coming. Buy Microsoft shares into any pullback.

10 Ways To Improve Cloud ERP With AI & Machine Learning

Capitalizing on new digital business models and the growth opportunities they provide are forcing companies to re-evaluate ERP’s role. Made inflexible by years of customization, legacy ERP systems aren’t delivering what digital business models need today to scale and grow. Legacy ERP systems were purpose-built to excel at production consistency first at the expense of flexibility and responsiveness to customers’ changing requirements. By taking a business case-based approach to integrating Artificial Intelligence (AI) and machine learning into their platforms, Cloud ERP providers can fill the gap legacy ERP systems can’t.

Closing Legacy ERP Gaps With Greater Intelligence And Insight

Companies need to be able to respond quickly to unexpected, unfamiliar and unforeseen dilemmas with smart decisions fast for new digital business models to succeed. That’s not possible today with legacy ERP systems. Legacy IT technology stacks and the ERP systems they are built on aren’t designed to deliver the data needed most.

That’s all changing fast. A clear, compelling business model and successful execution of its related strategies are what all successful Cloud ERP implementations share. Cloud ERP platforms and apps provide organizations the flexibility they need to prioritize growth plans over IT constraints. And many have taken an Application Programming Interface (API) approach to integrate with legacy ERP systems to gain the incremental data these systems provide. In today’s era of Cloud ERP, rip-and-replace isn’t as commonplace as reorganizing entire IT architectures for greater speed, scale, and customer transparency using cloud-first platforms.

New business models thrive when an ERP system is constantly learning. That’s one of the greatest gaps between what Cloud ERP platforms’ potential and where their legacy counterparts are today. Cloud platforms provide greater integration options and more flexibility to customize applications and improve usability which is one of the biggest drawbacks of legacy ERP systems. Designed to deliver results by providing AI- and machine learning insights, Cloud ERP platforms, and apps can rejuvenate ERP systems and their contributions to business growth.

The following are the 10 ways to improve Cloud ERP with AI and machine learning, bridging the information gap with legacy ERP systems:

  1. Cloud ERP platforms need to create and strengthen a self-learning knowledge system that orchestrates AI and machine learning from the shop floor to the top floor and across supplier networks. Having a cloud-based infrastructure that integrates core ERP Web Services, apps, and real-time monitoring to deliver a steady stream of data to AI and machine learning algorithms accelerates how quickly the entire system learns. The Cloud ERP platform integration roadmap needs to include APIs and Web Services to connect with the many suppliers and buyer systems outside the walls of a manufacturer while integrating with legacy ERP systems to aggregate and analyze the decades of data they have generated.

Boston Consulting Group, AI in The Factory of the Future, April 2018

  1. Virtual agents have the potential to redefine many areas of manufacturing operations, from pick-by-voice systems to advanced diagnostics. Apple’s Siri, Amazon’s Alexa, Google Voice, and Microsoft Cortana have the potential to be modified to streamline operations tasks and processes, bringing contextual guidance and direction to complex tasks. An example of one task virtual agents are being used for today is guiding production workers to select from the correct product bin as required by the Bill of Materials. Machinery manufacturers are piloting voice agents that can provide detailed work instructions that streamline configure-to-order and engineer-to-order production. Amazon has successfully partnered with automotive manufacturers and has the most design wins as of today. They could easily replicate this success with machinery manufacturers.

Company websites

  1. Design in the Internet of Things (IoT) support at the data structure level to realize quick wins as data collection pilots go live and scale. Cloud ERP platforms have the potential to capitalize on the massive data stream IoT devices are generating today by designing in support at the data structure level first. Providing IoT-based data to AI and machine learning apps continually will bridge the intelligence gap many companies face today as they pursue new business models. Capgemini has provided an analysis of IoT use cases shown below, highlighting how production asset maintenance and asset tracking are quick wins waiting to happen. Cloud ERP platforms can accelerate them by designing in IoT support.

Source: Capgemini Internet of Things (IoT) study, Unlocking the business value of IoT in operations

  1. AI and machine learning can provide insights into how Overall Equipment Effectiveness (OEE) can be improved that aren’t apparent today. Manufacturers will welcome the opportunity to have greater insights into how they can stabilize then normalize OEE performance across their shop floors. When a Cloud ERP platform serves as an always-learning knowledge system, real-time monitoring data from machinery and production assets provide much-needed insights into areas for improvement and what’s going well on the shop floor.

Industry Analysis

  1. Designing machine learning algorithms into track-and-traceability to predict which lots from which suppliers are most likely to be of the highest or lowest quality. Machine learning algorithms excel at finding patterns in diverse data sets by continually applying constraint-based algorithms. Suppliers vary widely in their quality and delivery schedule performance levels. Using machine learning, it’s possible to create a track-and-trace application that could indicate which lot from which supplier is the riskiest and those that are of exceptional quality as well.
  2. Cloud ERP providers need to pay attention to how they can help close the configuration gap that exists between PLM, CAD, ERP and CRM systems by using AI and machine learning. The most successful product configuration strategies rely on a single, lifecycle-based view of product configurations. They’re able to alleviate the conflicts between how engineering designs a product with CAD and PLM, how sales & marketing sell it with CRM, and how manufacturing builds it with an ERP system. AI and machine learning can enable configuration lifecycle management and avert lost time and sales, streamlining CPQ and product configuration strategies in the process.
  3. Improving demand forecasting accuracy and enabling better collaboration with suppliers based on insights from machine learning-based predictive models is attainable with higher quality data. By creating a self-learning knowledge system, Cloud ERP providers can vastly improve data latency rates that lead to higher forecast accuracy. Factoring in sales, marketing, and promotional programs further fine-tunes forecast accuracy.
  4. Reducing equipment breakdowns and increasing asset utilization by analyzing machine-level data to determine when a given part needs to be replaced. It’s possible to capture a steady stream of data on each machine’s health level using sensors equipped with an IP address. Cloud ERP providers have a great opportunity to capture machine-level data and use machine learning techniques to find patterns in production performance by using a production floor’s entire data set. This is especially important in process industries where machinery breakdowns lead to lost sales. Oil refineries are using machine learning models comprise more than 1,000 variables related to material input, output and process perimeters including weather conditions to estimate equipment failures.
  5. Implementing self-learning algorithms that use production incident reports to predict production problems on assembly lines needs to happen in Cloud ERP platforms. A local aircraft manufacturer is doing this today by using predictive modeling and machine learning to compare past incident reports. With legacy ERP systems these problems would have gone undetected and turned into production slowdowns or worse, the line having to stop.
  6. Improving product quality by having machine learning algorithms aggregate, analyze and continually learn from supplier inspection, quality control, Return Material Authorization (RMA) and product failure data. Cloud ERP platforms are in a unique position of being able to scale across the entire lifecycle of a product and capture quality data from the supplier to the customer. With legacy ERP systems manufacturers most often rely on an analysis of scrap materials by type or caused followed by RMAs. It’s time to get to the truth about why products fail, and machine learning can deliver the insights to get there.

Louis Columbus is an enterprise software strategist with expertise in analytics, cloud computing, CPQ, Customer Relationship Management (CRM), e-commerce and Enterprise Resource Planning (ERP).

I am currently serving as Principal, IQMS. Previous positions include product management at Ingram Cloud, product marketing at iBASEt, Plex Systems, senior analyst at AMR Research (now Gartner), marketing and business development at Cincom Systems, Ingram Micro, a SaaS start…


China Startups Brace For 'Capital Winter' As VC Funding Slows

Workers use computers at their desks inside a co-working space for start-up companies in Beijing, China. (Photo by Tomohiro Ohsumi/Bloomberg)

After a record amount of money gushed into China’s technology sector and fueled eye-popping valuations for many private companies in recent years, venture capital is said to be drying up and startups should brace themselves for what some industry insiders describe as a “capital winter.”

That’s the view held by several China-based venture capitalists and research firms. Speaking on the sidelines of Forbes Asia’s Under 30 Summit in Hong Kong, a number of young investors and honorees from this year’s list say valuations for private companies in China will return to “more reasonable” levels in the second half of this year. They said “bubble-like” valuation levels had become prevalent in China’s private investment space, as investors poured a staggering 1.2 trillion yuan ($180 billion)— or 1.5% of the country’s gross domestic product— to fund private firms in 2017, according to Beijing-based research firm Zero2IPO.

“There is too much money in China. The not qualified startups are getting funding, and the qualified startups are getting even more funding,” says William Zhao, vice president of Bertelsmann Asia Investments, a fund with more than $1.5 billion under management. “They are good companies, but are they worth that much? I think there is a bubble in it.”

More On ForbesChina To Account For A Quarter Of World’s Top 100 Venture Investors In Five Year

But those days appear to be coming to an end. As Beijing seeks to contain financial risks in the world’s second-largest economy, venture capitalists are having a much more difficult time raising money. This is mainly because authorities are clamping down on riskier investments made by Chinese banks, announcing in April a set of far-reaching rules on their wealth-management businesses. Proceeds raised through such channels have historically accounted for a sizable part of venture capital funding in China, as they were often later used to invest in private companies for higher returns. This can also lead to significant risks as the success of early stage firms is by no means guaranteed.

According to Zero2IPO, in the first three months of this year, venture capital and private equity firms in China raised 206 billion yuan ($31 billion), a 30% decrease from the same period last year. In a recent website post, Wang Ran, chief executive of Beijing-based investment firm CEC Capital, predicted that venture capital funding in China could fall by as much as 80% by year end. For private startups, this means valuation levels would be reduced by 30%, as there would be less money to support their growth, according to the post.

“It is indeed getting very hard for a lot of funds to raise money,” says Patrick Song, chief executive of CEC Data Capital, a fund affiliated with a subsidiary of the state-run China Electronics Corp. “Now, investors are much more cautious in their investment strategy, and valuation levels for many companies are sure to come down.”

Yellow Ofo bikes sit outside of the Park-n-Ride Alameda RTD Station on April 3, 2018 in Denver, Colorado. (Photo by Helen H. Richardson/The Denver Post via Getty Images)

One area that has been singled out for criticism is the so-called sharing economy. In a recent editorial, the state-run Xinhua News Agency said there is a “bubble” in this sector, with companies rushing to launch various product-sharing schemes that don’t have a viable path to profitability. It cited the example of home-sharing startup Zhubaijia, which is similar to Airbnb, but has been suffering from mounting losses that totaled 87 million yuan ($13 million) in 2016 due to fierce competition and high operating costs. In early July, regulators de-listed Zhubaijia from China’s third stock exchange because it failed to file its 2017 annual report on time.

And some of China’s better-known sharing-economy companies are scaling down their operations. After expanding into a dozen global cities in the past two years, Beijing-based bike-sharing startup Ofo laid off the majority of its workforce in the U.S. to re-focus on China. Meanwhile, its biggest rival, Mobike, was acquired in April for $2.7 billion by China’s largest online services platform Meituan. And Guangzhou Yueqi, a smaller player that operates the Xiaoming Bike brand, went bankrupt last year amid fierce competition that has pushed prices down to less than 1 yuan ($0.16) per hour in China. The company has more than 55 million yuan ($8.2 million) in outstanding debt after deploying more than 400,000 bikes across the country since 2016, according to Xinhua.

Meanwhile, as it gets harder to raise money from private investors, a record number of Chinese tech firms are tapping public markets. So far, several dozen Chinese tech firms, including Meituan and the three-year-old e-commerce startup Pinduoduo, have opted for initial public offerings. But the hotly anticipated IPO of smartphone maker Xiaomi valued the company at only about half of its initially proposed target of $100 billion, signaling that the capital markets may not support the lofty valuations set during earlier fundraising rounds for these tech companies.

More On ForbesIPO Of Chinese E-Commerce Firm Pinduoduo Mints New Young Billionaire

But this is not to say that China’s startup boom is over. According to CB Insights, the country was home to 55 unicorns, or private companies with a valuation of $1 billion or more, as of September 2017— a number second only to the U.S. And Ant Financial, the payment affiliate of e-commerce giant Alibaba, took over ride-sharing firm Uber as the world’s highest valued private startup after completing a $14 billion mega-funding round in June.

Moreover, investors seem to be especially optimistic of blockchain technology, the digital ledgers that underpin digital coin transactions. Although Beijing banned initial coin offerings last year to curb illegal fundraising activities, this technology of secured and decentralized data storage still holds a great deal of promises in China. It can be used in industries such as finance and real estate to facilitate faster and more efficient tracking of data and contracts, thereby improving efficiency. Investors are already making early bets now, predicting that blockchain will give birth to China’s next tech behemoth.

“Blockchain will be the next disruptive technology in China,” says Li Yao, partner at China’s Tsing Ventures, a venture capital firm affiliated with the country’s prestigious Tsinghua University. “It will take some time to be ready, but it definitely has a lot of potential in many different markets.”

More On ForbesThe Next Frontier For Billionaire Investor Jim Breyer: China And Blockchain

Cyber Saturday—Tenable IPO, Facebook Crash, Google Security Keys

Facebook crashes. Facebook shares lost nearly 25% of their value, wiping $120 billion off its balance sheet, after the company reported slowed user growth and weaker-than-expected revenue projections. The market plunge followed a host of privacy issues affecting the company—most notably, the blowback from Cambridge Analytica’s alleged misuse of people’s data and the enactment of GDPR, a new regulatory regime, in Europe. Meanwhile, BuzzFeed got a hold of a March memo penned by Facebook’s outgoing security chief, Alex Stamos. In it, Stamos lays out what he believes Facebook must do to fix its problems, such as reeling in “creepy” data collection practices.

Google gets serious. Google had a number of security updates this week. The search giant said it will start selling hardware security keys, branded “titan,” to corporate customers in an effort to combat phishing attacks and and hacking attempts. (Google uses such fobs to protect its own systems.) Google’s cloud business said it is adding a feature to help G Suite administrators identify virus-infected computers and malware-laden files. Also, Google’s Chrome browser has started labelling sites that don’t encrypt Internet traffic with “HTTPS” as “Not Secure.”

Life-unlocked. The website for Lifelock, an identity theft protection service now owned by cybersecurity giant Symantec, had a bug that exposed subscribers’ email addresses. After journalist Brian Krebs contacted Symantec about the issue, the company took the site offline. The member portal is back up, and Symantec says the issue has been addressed.

V for Vendetta. Online disputes can lead to grudges, which can can lead to bizarre, horrible consequences. This piece by Kashmir Hill at Gizmodo tells the story of how an argument about the decorum one should exhibit on the site of a former concentration camp spiraled into a life-wrecking nightmare. The Internet is a weird place—be careful out there.

Hey, literal jailbreaking!

Share today’s Data Sheet with a friend:

Looking for previous Data Sheets? Click here.

A Southwest Airlines Crew Member Just Asked Passengers to Mercilessly Mock United

Absurdly Driven looks at the world of business with a skeptical eye and a firmly rooted tongue in cheek.

At Southwest Airlines, employees are encouraged to express themselves.

To dance around in TV ads, for example.

Or to be funny. 

Here, then, was a Southwest flight that came upon a mysterious companion, flying parallel on its right.

This happens often on certain routes. 

Sometimes, it’s fascinating to see how long the planes will continue to travel side by side.

Sometimes, you might even wonder how close the other plane truly is.

In this case, a passenger was filming, as a Flight Attendant or pilot made an announcement:

Ladies and gentlemen on the right side of the aircraft, look over to your right. You’ll see United. Wave to them so they get to see what a full aircraft looks like.

Some passengers laughed, of course.

And it’s an excellent example of how Southwest can boast a strong brand while United’s is, well, in transition.

Some might think, though, that this wasn’t a gesture of solidarity toward the friendly skies.

I asked United what it thought of this barb. Stunningly, the airline declined to comment.

I also contacted Southwest to ask for its view.

“Our Employees like to have fun with our Customers. This sounds like some friendly banter with a competitor and friend of the skies,” a Southwest spokeswoman told me.

Friend of the skies. Ah, very good.

There’s largely little competition between the big four airlines.

Yes, they nibble at each other around the edges. They can even make barbs. 

Why, American Airlines CEO Doug Parker earlier this year referred to Southwest as “the cattle car.”

While I, of course, applaud Southwest for its readiness to be quippy, there’s one rational element that niggles here.

I’d much rather fly on an empty plane than a full one.

Indeed, research has regularly shown that the biggest difference between an enjoyable Economy Class experience and a less enjoyable one is whether the seat next to you is empty.

Then again, airlines have constructed things so cleverly these days that there seem very few empty seats anywhere.

Business Checkup: 4 Vital Signs To Watch

Given the depth of the customer/employee link, it’s important to consider how in touch you are with the systems that shape the experiences of both at your organization. Being intimately in tune with the people, processes and systems that run your organization are critical to the long-term performance and health of your company.

So, how do you know if you have a “healthy” company? Keep a constant watch on the vital signs. The following are diagnostics that will help you deduce what areas of your organization are compromised, and which are thriving.

Efficient collaboration is key

How efficiently different departments work together is a crucial indicator of your organization’s future performance. “Collaboration defines how your corporate mind collectively thinks ― a foundational component of its ability to innovate and compete,” said Matthew Leppanen, director of product and business productivity at Rogers Communications.

When analyzing the collaborative mindset at your organization, consider the following metrics:

  • Are misunderstandings causing slowdowns? Evaluate the time it takes to fulfill a product or service, then document communications protocols that will help teams reach productivity goals.
  • Are teams doubling down on work? Check for process redundancies between departments.
  • Are data silos muddying the waters? Inspect whether your talent is operating from a single source of data or is hampered by data silos, duplications, and outright errors. What you know about your customers can make the difference between success and failure. But only if you have good data, and only if your entire team is using the same sets.

NPS and eNPS: Golden insights

Your NPS (net promoter score) and eNPS (employee net promoter score) are critical insights into the well-being of your company, and they are usually directly linked to one another. As I mentioned earlier, unhappy employees often produce unhappy customers.

As Nick Mehta, Gainsight’s CEO, said in a presentation earlier this year, “Some people don’t realize the strong connection between making our teammates successful and our customers successful.” And it’s not just a matter of employee satisfaction, but also one of employee loyalty. Gainsight found employees who would strongly recommend their employers to others worked for organizations with:

  • 10% higher customer ratings
  • 21% higher productivity ratings, and
  • 22% higher profitability ratings

Keep a close eye on these two scores, and you’ll have a better handle on the pulse of both the customer journey and your viability in today’s talent war.

Can your organization deliver?

Another facet of business stamina directly impacted by your employees’ morale is the ability to fulfill products and services to customers on time. “Productivity and efficiency can be achieved only step-by-step with sustained hard work, relentless attention to details, and insistence on the highest standards of quality and performance,” said J.R.D. Tata, who oversaw India’s largest industrial empire.

In assessing the full scale of your operational productivity, map your company’s entire process from customer inquiry through fulfillment. Is your turnaround time consistent, or does it tend to take longer based on demand? Your NPS, customer surveys, and customer service requests and reports are excellent sources for this data.

Of course, production speed is not the only measurement when it comes to delivery of your products and services. The quality of what you distribute to your client base is just as important ― if not more so ― and is once again dependent on a harmonious workforce. Is your organization able to maintain quality standards as you grow?

Innovation: Survival of the fittest

When it comes to beating, or even keeping up with, the competition in a marketplace disrupted by digital transformation, staying ahead of the game is an imperative. If you don’t have plans today for what your company’s tech will look like in five years, you’re already behind the eight ball.

This is where a service blueprint ― a lean method for identifying the mechanics of your business and service experience ― can come into play. The process gives you a detailed look under the hood at the countless actors (customers, suppliers, partners, employees, etc.), touchpoints, and current systems and processes upholding your company’s infrastructure. From this invaluable insight, you can identify opportunities for innovation and where digital tools can be leveraged to streamline systems and processes, among other ways to improve the business.

Remember, in assessing your organization’s vital signs, the goal is to dig into the specific measurements — people, processes and systems — that indicate the state of the company’s essential functions. In many cases, a service blueprint can be just what the doctor ordered as it provides a 360 view of not just the customer journey and experience, but also the internal infrastructure needed to fulfill that experience.

Amazon's face ID tool mismatched 28 members of Congress to mug shots: ACLU

SAN FRANCISCO (Reuters) – A facial recognition tool that Inc (AMZN.O) sells to web developers wrongly identified 28 members of the U.S. Congress as police suspects, in a test conducted by the American Civil Liberties Union (ACLU), the organization said on Thursday.

FILE PHOTO:’s logo is seen at Amazon Japan’s office building in Tokyo, Japan, August 8, 2016. REUTERS/Kim Kyung-Hoon/File Photo

Amazon, in a response, said it took issue with the settings of its face ID tool during the test. The findings nonetheless highlight the risks that individuals could face if police use the technology in certain ways to catch criminals.

Since May, the ACLU and other civil rights groups have pressured Amazon to stop selling governments access to Rekognition, a powerful image ID software unveiled in 2016 by the company’s cloud-computing division.

The groups cited use of Rekognition by law enforcement in Oregon and Florida and warned that the tool could be used to target immigrants and people of color unfairly.

Their activism has kicked off a public debate. The president of Microsoft Corp (MSFT.O), Amazon’s rival which also uses facial recognition technology, called on Congress earlier this month to study possible regulations.

On Thursday, three Democrats who were identified in the ACLU test – Senator Edward Markey, Representative Luis Gutiérrez and Representative Mark DeSaulnier – sent a letter to Jeff Bezos, Amazon’s chief executive, expressing concern and inquiring about the tool’s accuracy and use by law enforcement.

Facial recognition is already widely used in China for police purposes, and a number of start-up companies there – some valued at billions of dollars – are aggressively pursuing the technology.

Amazon has touted a range of uses for Rekognition, from detecting offensive content online to identifying celebrities.

“We remain excited about how image and video analysis can be a driver for good in the world,” a spokeswoman for Amazon Web Services said in a statement, citing its help finding lost children and preventing crimes. She said Rekognition was normally used to narrow the field for human review, not to make final decisions.

The ACLU said it paid just $12.33 to have Amazon Rekognition compare official photos of every member of the U.S. House and Senate against a database of 25,000 public arrest photos.

The technology found “matches” for 28 members of Congress with at least 80 percent confidence, Amazon’s default setting, the ACLU said.

These matches were disproportionately people of color, according to the ACLU. Some 39 percent were African-American and Latino lawmakers, versus 20 percent who identify as a person of color in Congress, the ACLU said. It added that Rekognition could exacerbate harm because people of color are already targeted at an above-average rate by the police.

“Face surveillance is flawed, and it’s biased, and it’s dangerous,” Jacob Snow, an attorney at the ACLU of Northern California, told Reuters.

Amazon’s spokeswoman said the 80 percent confidence setting was appropriate for identifying objects, not individuals. The company guides customers to set a threshold of 95 percent or higher for law enforcement activities, she said.

Reporting by Jeffrey Dastin in San Francisco; Editing by Adrian Croft and Lisa Shumaker

A.I. Will Paint Your House: An Interview With Tanuja Singeetham

Tanuja SIngeetham is Vice President, Digital Marketing at Behr Corporation.Tanuja Singeetham

We interviewed Tanuja Singeetham, Vice President of Digital Marketing for Behr Corporation, and discussed how she leverages consumer insights and advanced technology, including artificial intelligence, to connect emotionally with Behr paint consumers and help them find their perfect color.  

Rosenthal: Tell us a little bit about yourself, work history, and your current focus at Behr.

Singeetham: I am currently the Vice president of Digital Marketing at Behr, and I’ve been here for the past four years. My areas of responsibility include our website, social media, mobile applications and I lead our e-commerce business on When it comes to digital advertising, I run all the search engine marketing efforts and work with our advertising team all of the display ads and campaigns.

Prior to coming to Behr, I spent 21 years at Nestle USA in Glendale, California. I actually started back in the day when, believe it or not, there was no email at Nestle! 21 years ago, when I got into digital, I launched one of the first websites for Nestle ever back in 1998; it was called and was one of the first content-focused portals for CPG, at a time when most brands were creating brand-focused websites. It was very early for its time and is something more common today in the CPG category.

Rosenthal: What is one digital initiative you’re proud of that you’ve worked on at Behr, and/or what about your job are you passionate about?

Singeetham: Coming to Behr was a huge opportunity! The organization had good foundational pieces: it had a website, an app and both have large audiences. Part of my decision to come here, was that there was an opportunity to advance the organization in digital. At the time, we had very little social media presence, and we weren’t really investing money in it.  We also had very little search or digital advertising, since we were mainly focused on traditional advertising, such as TV and print. A very small percent of the budget went to digital channels. Part of my goal when I started was to increase the knowledge and expertise internally and provide more guidance on how we could accelerate in the digital space.

The great thing about this category is that if you raise the conversation to something higher than what’s in the can, to instead focus on color and design, you have a very rich territory. Remember when you were a kid and you met another kid, one of the first questions you would ask was “what’s your favorite color?”. Also, many people’s first big decision about a space is when they’re a kid and their parent says, “What color do you want paint your room?” So coming into this category, I understood there are real emotional connections for consumers, and I wanted to bring that to Behr’s digital ecosystem.

With all the efforts we’ve made over the past four years, I am proud to say we have become leaders in social media for our category, grown our website traffic and engagement significantly, experienced triple-digit growth in ecommerce and have created a culture of data-driven decision making across all our online efforts.

Rosenthal: Can you share something about your upcoming digital initiatives at Behr? and/or what are the technologies that you’re most excited about over the next 2-3 years?

Singeetham: We conducted a research project to better understand paint in relationship to category adjacencies and overall home improvement projects.Through that research we learned the importance of feelings and emotions in the decision to take on a paint project.  Additionally we know through data that it takes consumers, in general, over a hundred days to narrow down to a color choice. Most of the time when people don’t buy the paint or move forward with the project, it is because they can’t make a decision on the color. It can be an overwhelming analysis process because there are just too many choices.

To help consumers on their color-decision journey, we launched a few things that have really been about applying technology to an insight based on real human behavior and individual consumer inputs. The first initiative, “Color Discovery” is a tool that allows consumers to go in and start with their project, and then pick what feeling they’re trying to achieve for their space. So now instead of “here’s a whole catalog of over 3000 colors and pick one” instead we say, “here’s a curated, personalized recommendation based on what they’re trying to do.” In fact we found that people who use this tool are much more likely to take a sample, than the average site visitor. It’s the best of what individual consumer data combined with advanced technology can do to deliver personalized recommendations.

Second, we launched a tool on our site called “Pins to Palettes,” which leverages a key touchpoint along the inspiration part of the consumer journey, Pinterest. Consumers go to our site and log in to their Pinterest account, and then select up to six pins across multiple boards related to their project.  Using our patented technology, we scan the images and are able to give them a coordinating color palette across the various items personalized for them.

In both cases we make it easy for consumers to try on the recommended colors virtually with our Paint Your Place tool, share colors and photos via email or social network, and make it easy to buy a sample at Home Depot so they can get started on their project.

Rosenthal: Messaging delivers on the promise of personalization. How does personalization change the way you think about the lifecycle of the consumer? In the near future, what does messaging look like between a customer and a brand?

Singeetham: As we move forward, we’re thinking about some of the other big things that have shifted in the world. One major one is the “concierge movement”. It’s only continuing to grow. First, we launched live chat color recommendations. People can go onto our site and actually talk to customer care reps and get advice and links back. We’ve seen very positive scores on our consumer satisfaction. It’s up there with the highest of the industry.

So, then we took it into the whole idea of giving recommendations in our color clinics. We work with influencers and our internal color experts and we launch events where we go inside of the social world and allow consumers to ask their questions right on the on Facebook, Twitter or Instagram, and then we’re able to give them immediate assistance. We have a command center of people to provide specific color recommendations based on whatever the consumer is looking for.

Rosenthal: Unlike social media, which is one to many, messaging is one-to-one marketing at scale. When you think about the promise of the messaging space, why are you excited about this space?

Singeetham: Trying to move personalization beyond live chat can be challenging to scale, as consumers are online 24/7/365 days a year. That’s why our chatbot exploration is an amazing opportunity!

We have so many learnings on the types of questions consumers are asking and know they are looking for answers. We’re doing significant exploration on natural language processing and have a great opportunity to find ways that we can leverage that and provide 24 hour service in this recommendations area. That is one of the things that we are exploring with Snaps, and we are working also with a couple of other AI providers in other areas of artificial intelligence.

Rosenthal: Is Behr looking at Artificial Intelligence or machine learning elsewhere in their business?  

Singeetham: We currently working with IBM Watson, as well as Google and their cloud team, on various projects where they are analyzing data. We use this data to identify clusters and gain insights to be smarter and more effective with our marketing communications. For example, we may see that people in the Northeast tend to go for grays or neutral living rooms. So instead of just doing generic display ads and banners, we could customize images to be more relevant to an audience and be more meaningful with our outgoing messages. It allows us to deliver personalization at scale, which is what this category and the marketing industry as a whole, is headed toward.   

Samsung's Galaxy Note 9 Is Extremely Expensive

This could be the end for the Galaxy Note. Following the revelation, Samsung plans to cancel the range if Galaxy Note 9 sales disappoint, new information from a rock-solid source suggests the (at times controversial) model now has little chance of succeeding… 

WinFuture’s Roland Quandt, who has a near-flawless track record and most recently exposed Google’s Pixel 3XL, has confirmed Samsung will jack up the asking price for the Galaxy Note 9 to potentially unrealistic levels.

Galaxy Note 9 concept proved too

Quandt says Samsung will start Galaxy Note 9 pricing in Europe at €1050 ($1230) for a 128GB model with a flagship 512GB option costing €1250 ($1460). By comparison, last year the Galaxy Note 8 launched at under €900. If you’re looking a consolation, the Note 8 only had 64GB of storage.

Furthermore, were Quandt’s track record were not enough, this data also ties in with a leak from Samsung sources last week.

It is worth noting at this point, smartphones are more expensive in Europe but mostly because their prices always sales tax. As such, the Galaxy Note 9 looks primed to launch Stateside with prices of $999 and $1199, excluding tax. That won’t hack it in a year when Apple will actually cut iPhone prices substantially.

Galaxy S10 is where Samsung’s ambition lies (concept)Concept Creator

In defence of the Galaxy Note 9, it does have subtle but sizeable upgrades for which Samsung deserves credit. But when its biggest feature was pulled, that proved Samsung’s focus instead lay with its upcoming 10th anniversary Galaxy S10.

Consequently, laying the future of the Note range at the feet of the Galaxy Note 9 in its most incremental of years seems grossly unfair. It reeks of a company merely looking to rubber-stamp a decision that was apparently made long ago.

Unlike the vast majority of smartphones, the Galaxy Note series is unique with its focus on practicality and productivity. With Samsung’s next big phone innovation set to cost $2,000, it’s sad to see the Galaxy Note 9 (almost inevitably) end the range with a whimper, not a roar…


Follow Gordon on Twitter, Facebook and Google+

More On Forbes

Samsung’s Galaxy Note 9 Is Very Expensive

Samsung’s Galaxy S10 Is A Truly Massive Smartphone

Massive Galaxy Note 9 Leak Details All-New Features

Samsung Filing Makes Galaxy Note 9 Official

Samsung’s Radical Galaxy Smartphone Costs $2,000