Comment by michaelochurch | Hacker News Reader

Comment by michaelochurch | Hacker News Reader

He's conflating an industry with a geographic location. In the 1960s, "Silicon Valley" was dominated by big companies. Now it's overrun with dipshit social media concerns that won't last 10 years. It was a different thing altogether in 1975 and in 1999. In 2030, two questions are:

* What will be the geographic center of innovation?

* What industry will the innovation be in?

These are orthogonal. I can't answer either of these. I honestly have no clue. I'll make one solid prediction, though:

I don't know what the Big Thing of 2030 will be, but it will involve computation. It'll be far outside of traditional IT (the use of computers to implement 20th-century business ideas) and social media (same, but for 21st-century cultural experiments) but whatever is Big during 2030 will require a lot of skill when it comes to computation.

Of course, I have no idea whether this will happen in Northern California or some other corner of the world.

What I think is clear is that VC-istan innovation is over. The problem is that the VCs talk to each other and collectively decide who's hot and who's not. So, instead of getting true risk-taking, you have selection by committee in the funding process. That is starting to fail.

"VC-istan innovation is over"

Maybe. Much as I'd like to believe it, there are some things that can't be bootstrapped. Actually there are many things. We just don't do them because the VC's aren't funding them either. Perhaps they'll go back to it.

OTOH, VC access is one big reason innovation gets centralized to places like Silicon Valley. Reduce that friction and it can spread out more. Still doesn't fix the issue of networking for talent which is another network effect that clumps innovation.

I did my first startup in Houston, Texas. It was great. Extremely cheap. Built a product and got to profitability on about $600K of capital. Moved the company to Silicon Valley before ultimately selling it to Borland (it became Quattro Pro).

Houston worked for development because I could get talent out of Rice University and Univ. of Houston. Infrastructure was cheap because they were having a real estate crunch at the time. You could get free rent for 2 years in exchange for signing a 5 year lease.

What was hard was the marketing and networking. After riding planes to go to the East and West Coasts to get the product reviewed, I decided we had to move the company.

In the Internet age, that may be a lot less true. No particular advantage to doing my current bootstrap (CNCCookbook) in Silicon Valley other than that's where I live. Houston would be a lot cheaper.

Much as I'd like to believe it, there are some things that can't be bootstrapped. Actually there are many things. We just don't do them because the VC's aren't funding them either. Perhaps they'll go back to it.

Funding is the art of boss-picking. Who gets to be a boss, and who has to implement someone else's idea and have a boss? Funding goes to a (Person, Idea) pair-- a 2-tuple, in other words.

Now, the problem. Person-picking is fairly useless. It's not that good management isn't important, but there's no correlation between being fundable (i.e. attractive to VCs, probably because one went to the same boarding school) and being a good executive. Idea-picking is important, but very hard to do.

Now, there are obvious Ideas that deserve funding. Cancer research, clean energy. The issue is that most of us who are intellectually honest know that we don't make a credible (Person, Idea) pair with the meatiest Ideas (because others know a lot more than we do). The higher the quality of Idea, the fewer Persons there are for whom there's a credible pair. (This social-media nonsense is mostly low-quality ideas that a 5-year-old could execute.)

So, what do you have in VC-istan? You have narcissists who say, "I don't care what the Idea is, but I deserve to be the Person because, goddamn it, I have an MBA." They continually pivot until they find something a VC will fund. Typically, they land on Ideas for which it's easy to have a (Person, Idea) match, which tend to be those that involve no vision or real technical meat except for "scaling", which isn't that hard if you're half-decent at hiring technical people.

Regarding computation, I respectfully disagree. I believe the shift from be from software back to hardware again.

Failures in Computation: 1) general artificial intelligence, this has been touted that it's just around the corner since 1970's; so far most practical AI has been just variants of expert systems. 2) Bioinformatics, this was touted in the early 2000's and by the Kurzeweil-crowd as the singularity moment that will bring advances to biotech; unfortunately, all of the mountains of data has not yielded any concrete results because it's too complex for simple statistical methods. 3) Quantitative finance: from the failure of stat arb funds in 2007 to the blind trust in BSM in derivatives risk management software, mathematical models have beautiful assumptions that the world is smooth whereas the real world is not.

Emergence of Hardware: Fab labs, 3D printers will enable delivery of real consumer products via CAD schema. Alternative energy research, medical robotics for remote surgical operations. Just look at the real trailblazers of Silicon Valley, SpaceX, Tesla and Google Driver-less cars. All hardward projects. Personally, I'd like to bone up EE, molecular biology lab protocols, CAD, MechE skills instead of RoR or Facebook APIz.

I would assume driver-less cars are primarily a software achievement. Same with robotic surgery, but I'm not sure.

Also, you can't compare two things by comparing the failures of one to the successes of the other.

I think AI has achieved quite a lot. Consider Google and Amazon, much less Watson. AI is like "magic". Whenever we figure out how to do something, it's no longer AI. It's statistical learning or computer vision. However, the problem of general AI turned out to be a lot harder than we thought it would be. Moravec's Paradox deserves mention here.

Hardware and software will interact more in the 2010s to '30s. What will people design all this awesome hardware with? Software. Taking the other direction, I do agree that the hardware world is becoming more interesting to people who might, a generation ago, have thought of themselves as "pure software" people. In 1985, you could "do AI" and never go outside of a garbage-collected language (Lisp). In 2013, it's hard to do high-performance machine learning without having some knowledge of all the neat things people can do with hardware (e.g. GPU programming). Also, the existence of continuing education options like Coursera is making people more Y-shaped than T- or I-shaped.

I agree with you that a lot of the "hot" work in the social media world (learning a bunch of damn APIs) isn't very hard, deep, or interesting. However, don't rule out software entirely.

I understand T-shaped, but what are Y-shaped and I-shaped?

I-shaped: depth with no breadth (highly specialized, no cross-disciplinary knowledge).

Y-shaped: some breadth at intermediate levels. More depth in one's non-specialty areas than T-shaped. (The top of the T becomes a triangular funnel instead of a bar.)

Thanks!

One thing I've noticed is that many of the most successful people are multi-specialists, i.e., they have two (or more) "specialties" that combine in a rare and valuable way. Examples include the rise of the "growth hacker" or "data scientist", who are typically programmers applying their craft towards marketing or statistics respectively.

There's both a perception component and a real value component. If you're the best Marketer, other Marketers will frequently deny it, get jealous, or sabotage you. But if you're "the Marketer who also writes code", nobody feels threatened. You're competing in a different area, so people can support you and ask you for favors when they need help.

In terms of real value, a junior programmer with strong business skills in a company with no other programmers can contribute millions of dollars of value through basic automation-I've actually done this. The same programmer will have a much lower dollar value when surrounded by better programmers, at least initially, because all the low-hanging fruit has been taken.

Excellent observation. This relates to my observation that it's better to be an "X Who Programs" than a "J.A.P." (Just A Programmer) in most cases. Unfortunately, corporate HR systems are built on the idea that people need to be reduced to a single, specific role.

Dimensionality is the rule, here. We suffer from it in hiring, where it's literally impossible to meet perfectly the requirement lists many companies put out (demanding 5 years of experience in a 3-year-old NoSQL database). This seems to be a part of why hiring is so hard on both sides.

Yet, once people are working together, if people aren't careful, they forget dimensionality altogether and attempt to form a total order, and it's this process that leads to the insecurity and sabotage.

He's conflating an industry with a geographic location.

http://en.wikipedia.org/wiki/Metonymy

Fair point. I guess my issue was that Silicon Valley != VC-istan, the latter being just one incarnation of it out of many that have existed. VC-istan could die and there'd still be something that would emerge to make use of that talent (maybe old-style R&D could come back; unlikely, but one can hope).

>overrun with dipshit social media concerns that won't last 10 years

actually, the most innovative companies today are in the social media space, on account of them being able to attract the most talented developers. ofcourse, what they do today on the outside may not be that innovative, but what goes on inside a facebook, twitter, google, linkedin, is certainly light years ahead of what goes on inside investment banks & bread-and-butter techshops ( my past employers ). you are assuming social media outfits will be forever doing social media, which is 100% wrong. social media is lucrative & network effects are huge => large money inflow => R&D into non-social-media products. Very similar to what Ma Bell did by overcharging everybody on phone calls & routing all the cash to Bell Labs R&D. There are entire new languages & systems for handling big data that were invented within social media companies. Just because the dataset for those is social content does not mean that's all they can handle. Social media companies, atleast the big 4, are definitely primed for handling cheap large scale computation on a massive scale, the exact problem that you think will dominate in 2030.

I don't put the companies you listed in the "dipshit social media" category. Facebook and Twitter had a vision other than "get acquired" behind them. The "Me, too" players focused on M&A are the dipshits.

Reply on news.ycombinator.com