Sunday, December 11, 2016

If it walks like a duck, and quacks like a duck...

Hot on the heels on Theranos, Hampton Creek, and other companies reputed to be vastly exaggerating (or downright lying) about the capabilities of their technology, Magic Leap is now under scrutiny. This company promises Virtual/Augmented reality and is a darling of VCs, with technology described as far ahead of the competitors, fawning pieces in international tech publications, big name investors, and a monster multi-billion dollar valuation (based on the $1.4 billion of funding so far) - all without an actual product or public demonstration. That, if you've been reading this blog, is the start of a familiar story.


There have been questions over the actual state of Magic Leap's technology for some time, but this week saw an article by The Information which details not just these questions over technology, but that the videos of the technology in action are not truly representative of the reality, despite claims by the company to the contrary. It's one thing to raise significant money based on an audacious goal of where you can take technology - hardware is *hard* and needs a lot of time and work to bring to fruition - but if you pass off fabricated demonstrations as the state of the actual tech (rather than as examples of what could be) in your fundraising, then that's fraud, and at least one federal agency will be taking a keen interest in knowing more.

As I've noted before in blogs, there's a tremendous amount of pressure on startup founders to cast their company in the best possible light, and it's easy to let that slip from "best possible" to "exaggerate" to "downright lie". Investors encourage this behaviour by repeatedly funding the more egregious cases, and ignoring those who are more honest and conservative in their goals.

I simply don't know enough about Magic Leap to say, one way or another, if they have what they claim - it's not an Energous or similar company where you can prove with maths and physics that their claims are not realizable - but if what they showed investors was the same as they claim publicly, a lot of that $1.4 billion is going to end up spent on legal fees. 

Since release of this article by The Information, the company seems to have done everything except provide actual proof that their claims are real. They've fallen back on talking about how hard they are working, and that great things are coming, but nothing solid (sound familiar?). I'm just waiting for the seemingly obligatory "Haters gonna hate", "Disgruntled ex-employee", and "It's just the sexist press and they wouldn't be doing that if the CEO was a man" (ooops, sorry, wrong company). The Register details the company response here, in a very entertaining article I encourage you to read.

Now you might comment that the funding for Magic Leap comes from private sources, that VCs and investors like Google are capable of looking after themselves, and it's not our business what they do with it. To an extent, that's true, however it covers a major problem we face in the misallocation of resources. $1.4 billion is not a trivial sum, and represents not only resources that could be applied to something of real and actual value (roughly 500 good engineers and support staff working for 10 years), but creates the standard by which all other companies now must be compared. Imagine you've a small company with solid VR technology that actually works and can be delivered as a product, but when you present it to a VC you're told "Magic Leap already beats that - I won't invest, there's no market". Because you are honest, you don't get funding and your company never takes off, we as the public don't get the benefit of that technology, and the VC's investors (like pension funds) don't get the benefit of the profits. Worse, it encourages the less-than-honest founder to "exaggerate" capabilities and exacerbates the problem. As a society we all lose from this.

We'll have to see how this one turns out, I had hoped Magic Leap were onto something, but the reaction of the company does not encourage me. Now I had been looking to write a blog on "Guide to Tech Companies with Exaggerated/Nonexistent Tech" but The Register beat me to it in their Magic Leap article, with a checklist I just can't improve on.
  • Refusing to give a launch date.
  • Refusing to talk about the tech, claiming confidentiality or trade secrets.
  • Using news of investments or hires as evidence of technological progress.
  • Promoting itself on a big stage rather than in a small room.
  • Offering a well-crafted message and vision but becoming immediately vague when pushed on actual details.
  • Offering "exclusive access" – with restrictions.
  • Confusing working hard with making progress.

Can anyone think of other companies that tick off everything on this list?

Edit: The CEO did respond in a blog post on the company website and indicates that they are in the process of building building product equivalent devices, which is more positive than what had been written before. Still the question arises, will their product deliver what's been shown in the videos, or will it merely match what their competitors, like the Hololens or Oculus Rift, already ship?

For a slightly more technical analysis, this blog has some good info. Summary: What they show in the videos needs large equipment/computation that is too expensive for consumer products, so the actual released product will likely not match the videos, but be closer to that of the competitors.

Further update: Just keeping everything together in one post, rather than adding to a new one. I've had discussion with a few people over the last few days who've better knowledge than me on this, and from what I can piece together, ML do have some solid AR tech, however what you see in the videos is from larger scale equipment with powerful computers running it. The wearable consumer gear is highly unlikely to be doing this, yet, and don't be surprised if a first product isn't so amazing. There's no reason to think they won't deliver a product, but nothing to indicate they are massively ahead of their competition either. I expect this field will see the 'lead' pass back and forth between companies over the next few years, and also some market differentiation - which is a great thing for encouraging better performance and prices.

2 comments:

  1. First of all very warm thanks for your refreshing blog posts in a desperately grim landscape of tech journalism, from another engineer distressed by the way the majority of the online media are gradually losing touch with reality, living on self-perpetuating hype and hyperbole rather than on facts and nuance as their job should be.

    Secondly, with my apologies for hijacking this post's subject, and because you already addressed Tesla’s problematic approach to self-driving cars in an earlier post, I was wondering whether would care to comment on the recent developments at Tesla, with Musk's statements promising "full self-driving" cars within a year if not even earlier, only a few weeks after replying to a journalist that being skeptical of self-driving cars was tantamount to "killing people" ? I am shocked, by the way, that I couldn't find any proper criticism of what appears to me as an outright, 1984-style threat to freedom of thought and expression.

    For my part here's what I predict will happen in the next years: after having started to sell their cars with a bunch of cameras, a radar and some kind of tailored on-board computer people made much about – “full self-driving hardware” they pompously call it, but that’s just the easy part of the whole business, with reliable self-driving algorithms being magnitudes of order harder - Tesla will release ever more "software updates" to their cars (doesn't this sentence alone sound utterly wrong - software updates on life-threatening devices ?), ever reducing the need for driver intervention - but not suppressing it. The 1% of cases their software can't handle will be the hardest to get rid of...

    ...but here's the catch: as laid out convincingly in a thorough Guardian article (https://www.theguardian.com/technology/2016/oct/11/crash-how-computers-are-setting-us-up-disaster), it will be good enough for people to trust it dangerously, and start looking at their smartphones and doing other things.

    And then, people will be killed.

    But that's too bad for them, Mr. Musk will say, because the fine print in Tesla's terms - "driver should keep their hands close to the wheel and eyes on the road at all times" – will not go away any soon. Which is precisely why half-baked self-driving cars are useless: what’s the point of not driving if one must monitor the machine doing the driving the whole time ?

    ReplyDelete
    Replies
    1. Thanks for your comments. I generally agree with you, and deciding how much control to give systems in charge of many functions of our lives, and at what time, is going to be a problem for the next few decades at least.

      More what concerns me right now is how dumb humans are in what they allow. Tesla are saints compared to what Uber have just done. I'll be trying to post on this too - just struggling to get the free time. Hopefully Christmas break will give me that.

      https://www.theguardian.com/technology/2016/dec/19/uber-self-driving-cars-bike-lanes-safety-san-francisco

      Delete