3 ms·

Central to their reasoning is the thesis that every ~15 years there is a breakthrough in computing (PCs in ~'78, the Web in ~'93, Smartphones in ~'07). This the

Central to their reasoning is the thesis that every ~15 years there is a breakthrough in computing (PCs in ~'78, the Web in ~'93, Smartphones in ~'07). This thesis is laid out on their site. So, the hedge seems to concern investing in a new computing platform that will gain traction around 2021-2023. Does either the thesis or hedge seem bad? Well, the thesis doesn't seem obviously bad in comparison to other VC theses out there (I mean, theses are sort of about putting a stake in the ground with regards to predicting the future). Let's agree the thesis is sound. Is the hedge still bad? I don't know. If you were to line up the other options, like A.I. and AR/MR/VR (are there any others?), I'm not sure one would really stand out above all the others. Maybe A.I.? But even then, they've already invested in the other options (especially A.I.), so at this point it's a matter of distribution of invested capital. Could the crypto investments end up being complete and embarrassing failures. Definitely. But, I'm not sure how this risk really stacks up to a different hedge for 2021-2023. It doesn't seem to be obviously bad in comparison to the other options.

rsynnott 6y ago
So, wait, their thesis is _numerology_? And numerology with suspect numbers at that; 1978 was the year after the introduction of the first viable personal computers, 1993 was, er, Mosaic came out I suppose, but it was not the year the web showed up (that was earlier) or the year that it started to see widespread adoption (that'd be, er, 95 or 96). 2007 was the year that smartphones became mainstream, if you don't count blackberries or other consumer quasi-smartphones. This all feels very cherrypicked to fit the thesis.
- api 6y ago
  Smartphones became popular in Asia long before the USA too, and there was a big market for flip phone apps there too while those never caught on in the US. This model seems more than even US-centric or Anglo-centric. Its Valley-centric. I don't know what their reasoning is, but I suspect it's a mix of everything discussed on this thread: ancap/right-libertarian ideology, an affinity for sexy tech regardless of usefulness, faulty reasoning by historical analogy, reality distortion due to having made money off the crypto bubble, and a bit of sunk cost fallacy. I am willing to be proven wrong. Show me growing uses of cryptocurrency that are not one of: niche wire transfer applications, black/grey market commerce, cryptocurrency aficionados nerding out, gambling, speculation, fraud, or solutions in search of a problem. Those are all (except the last two) real applications to varying degrees, but they're not large, rapidly growing, mainstream, or of that much interest to business customers. In other words show me something someone that doesn't care about cryptocurrency would use. The Internet in the mid-90s offered a ton of things people wanted to use even if those people didn't know what an IP packet was. I have asked the above in various forums, Twitter, etc. for years and so far nobody can produce anything. The closest I've ever seen is this: https://sia.tech https://sia.tech It's the only ICO project I know of that has actually shipped something usable and possibly useful. Still that's one singular example and it's not like they're going to displace Amazon S3 or Backblaze any time soon.
  - arcticbull 6y ago
    When your support page includes a whole section devoted to SEC Settlements you're not off to a rip-roaring start.
- anongraddebt 6y ago
  VC theses are almost always just pattern matching. Moreover, they aren't the first to ever put forward that thesis. Is it a good thesis? What's your criteria? By any scientific criteria, of course, the thesis is asinine. However, if the criteria is other VC theses, then it is an adequate thesis.
  - rsynnott 6y ago
    So... compared to other nonsense it passes muster?
    - anongraddebt 6y ago
      Why wouldn't it? Are there grades of reasonableness when trying to predict the future?
arcticbull 6y ago
The difference is AI and MR/AR/VR have use cases that aren't solved as well with classical methods. There isn't a single use case blockchain is better at than classical methods. Not one -- except for crime and heating up the planet. We've spent 12 years trying to find one. "AI" is great at solving fuzzy matching and pattern inference, which opens up a world of interesting things. Image segmentation, image analysis, dynamic image infill, facial recognition, and so on can all benefit. Is it a be all and end all technology? Probably not, certainly not yet, but it's clearly, trivially and obviously superior for a certain problem domain. Just because something happens "every 15 years on average" doesn't mean it happens on a 15 year clock and it certainly doesn't mean you can just throw a lawn dart and hit one in the face without using some logic, reasoning and due diligence.

rsynnott 6y ago

So, wait, their thesis is _numerology_? And numerology with suspect numbers at that; 1978 was the year after the introduction of the first viable personal computers, 1993 was, er, Mosaic came out I suppose, but it was not the year the web showed up (that was earlier) or the year that it started to see widespread adoption (that'd be, er, 95 or 96). 2007 was the year that smartphones became mainstream, if you don't count blackberries or other consumer quasi-smartphones. This all feels very cherrypicked to fit the thesis.

api 6y ago
Smartphones became popular in Asia long before the USA too, and there was a big market for flip phone apps there too while those never caught on in the US. This model seems more than even US-centric or Anglo-centric. Its Valley-centric. I don't know what their reasoning is, but I suspect it's a mix of everything discussed on this thread: ancap/right-libertarian ideology, an affinity for sexy tech regardless of usefulness, faulty reasoning by historical analogy, reality distortion due to having made money off the crypto bubble, and a bit of sunk cost fallacy. I am willing to be proven wrong. Show me growing uses of cryptocurrency that are not one of: niche wire transfer applications, black/grey market commerce, cryptocurrency aficionados nerding out, gambling, speculation, fraud, or solutions in search of a problem. Those are all (except the last two) real applications to varying degrees, but they're not large, rapidly growing, mainstream, or of that much interest to business customers. In other words show me something someone that doesn't care about cryptocurrency would use. The Internet in the mid-90s offered a ton of things people wanted to use even if those people didn't know what an IP packet was. I have asked the above in various forums, Twitter, etc. for years and so far nobody can produce anything. The closest I've ever seen is this: https://sia.tech https://sia.tech It's the only ICO project I know of that has actually shipped something usable and possibly useful. Still that's one singular example and it's not like they're going to displace Amazon S3 or Backblaze any time soon.
- arcticbull 6y ago
  When your support page includes a whole section devoted to SEC Settlements you're not off to a rip-roaring start.
anongraddebt 6y ago
VC theses are almost always just pattern matching. Moreover, they aren't the first to ever put forward that thesis. Is it a good thesis? What's your criteria? By any scientific criteria, of course, the thesis is asinine. However, if the criteria is other VC theses, then it is an adequate thesis.
- rsynnott 6y ago
  So... compared to other nonsense it passes muster?
  - anongraddebt 6y ago
    Why wouldn't it? Are there grades of reasonableness when trying to predict the future?

arcticbull 6y ago

The difference is AI and MR/AR/VR have use cases that aren't solved as well with classical methods. There isn't a single use case blockchain is better at than classical methods. Not one -- except for crime and heating up the planet. We've spent 12 years trying to find one. "AI" is great at solving fuzzy matching and pattern inference, which opens up a world of interesting things. Image segmentation, image analysis, dynamic image infill, facial recognition, and so on can all benefit. Is it a be all and end all technology? Probably not, certainly not yet, but it's clearly, trivially and obviously superior for a certain problem domain. Just because something happens "every 15 years on average" doesn't mean it happens on a 15 year clock and it certainly doesn't mean you can just throw a lawn dart and hit one in the face without using some logic, reasoning and due diligence.