Predictions are fun! Here I am keeping track of my predictions and a few of my friends’, trying to be reasonably specific about what exactly is being predicted and when it will happen. I take this as a game and won’t feel too disappointed if my own predictions don’t happen.

My pending predictions

PREDICTION-0016: 2016-01-07: Timeline for summoning a self-driving car around the Bay Area

  • From: @ebruchez
  • Text: “In 10 years, we will be able to summon an autonomous car with a mobile app to go to random addresses around the San Francisco Bay Area.”
  • Comments:
    • This implies some companies will provide that service (a la Uber).
    • There is no requirement to be able to go outside of the Bay Area (say Tahoe, etc.).
    • The system must be useful: that is, there have to be reasonable uses for people like us. For example, if there are only a couple of available destination addresses, the system will not be usable. The expectation is that most, but not necessarily all, destinations should be available. For example, taking you from your home to a friend’s home, or to a train station, etc.
    • The system must be usable also in that it must take a reasonable amount of time. It cannot be, for example, more than 50% slower than driving on your own in the same circumstances.
  • Resolution:
    • WAIT until 2026-01-07

PREDICTION-0003: 2013-09-26: Smart glasses technology

  • From: @ebruchez
  • Text: “In 10 years, glasses with features identical to 2013 Google Glass (or better) will be indistinguishable from regular prescription eye glasses (except if social or legal requirements impose some identifying aspects)”
  • Comments:
    • The main idea is that the technology for this will be there and that products will be on the market with that technology.
  • Resolution:
    • WAIT until 2023-09-26

Other people’s pending predictions

PREDICTION-0018: 2014-10-15: Driverless cars by 2023

  • From: @elonmusk
  • Text: “We’ll be able to achieve true autonomous driving, where you could literally get in the car, go to sleep and wake up at your destination.”
  • Comments:
  • Resolution:
    • WAIT until 2024-01-01

PREDICTION-0017: 2016-01-07: Timeline for summoning a self-driving car around the Bay Area

  • From: @avernet
  • Text: “In 15 years, we will be able to summon an autonomous car with a mobile app to go to random addresses around the San Francisco Bay Area.”
  • Comments:
    • Same conditions as PREDICTION-0016 but with a timeline of 15 years instead of 10.
  • Resolution:
    • WAIT until 2031-01-07

PREDICTION-0015: 2016-01-06: Timeline for summoning a self-driving car for a trip out of town on a rainy day

  • From: @avernet
  • Text: “In 10 years, we still won’t be able to summon an autonomous car, on a rainy day, to go from San Carlos to Healdsburg, with the trip taking a reasonable amount of time.”
  • Comments:
    • The “rainy day” is designed to make it clear that with a private car, the technical ability to do the trip would not be dependent on whether it is raining or not.
    • The “reasonable amount of time” qualifier is there to insist on the usability of the system: if the car takes the whole day to go there, it won’t be usable and therefore probably not used. As of 2016-08-03, that trip takes, by car just over 2 hours. We could say that if the trip takes, say, more than 50% of the time for the same trip by private car, the solution would be unacceptable. We can debate the exact percentage (50%, 100%, etc.).
  • Resolution:
    • WAIT until 2026-01-05

PREDICTION-0001: 2013-07-20: Death of Apple

  • From: @otazi:
  • Text: “Apple will be dead in 10 years.”
  • Comments:
    • What will “dead” mean? I will interpret it not as “out of business” as that seems impossible for a company of this size, but as “largely irrelevant”. This means significantly less relevant than in 2013. How does this translate in terms of market share or number of devices or services? Maybe we can set the bar at 5% market share, as suggested by @avernet.
  • Resolution:
    • WAIT until 2023-07-20

Completed predictions

PREDICTION-0025: 2020-11-10: Support for 32 GB of RAM in Apple’s ARM-based 13” MacBook Pro

  • From: @ebruchez
  • Text: “Within 6 months, Apple will add the option of having 32 GB of RAM in the 13 MacBook Pro.”
  • Comments
    • Today, Apple released their first Macs with Apple Silicon. But none of the existing machines support more than 16 GB of RAM.
    • I suspect that this is due to the availability of memory chips to reach 32 GB. I don’t think this means that there won’t be any larger configurations.
    • So here I am predicting that 32 GB of RAM (at least) will happen within 6 months.
  • Resolution:
    • FAIL: This is the case only since October 18, 2021, when Apple announced laptops with M1 Pro and M1 Max CPUs. The M1 Pro supports 32 FB of RAM, while the M1 Max supports 64 GB. I was a little over 5 months off in my prediction.

PREDICTION-0024: 2020-11-07: 2020 Presidential election aftermath and transition of power

  • From: @ebruchez
  • Text: “Following Joe Biden’s win, and in spite of litigation and posturing from Trump, the transition of power from Trump to Biden will be smooth.”
  • Comments:
    • It is easy to come up with Doomsday scenarios: Trump might never concede, electors might not cast their votes as usual, the Supreme Court might have to intervene, or the congress, etc. In addition, Trump could make the transition at the White House hell in other ways.
    • It wouldn’t surprise anyone if Trump did not receive Biden at the White House, as is customary, or if he didn’t shake hands with his successor. But this would not matter for anything consequential.
    • However, I think that overall there won’t be any catastrophic scenario. Trump will make some noise, then go and be out of politics. The most he will be able to do after that will be to Tweet.
    • Most of his supporters in politics will quickly move away from him. I expect he will keep a following on Twitter and the likes for a while, but that again won’t impact the transition to the new president.
  • Resolution:
    • FAIL: I am counting this as a failure. Trump never conceded his loss and repeated he would never concede even on January 6, 2021. He incited the subsequent storming of the Capitol, which caused deaths and could have turned way worse. He behaved in ways that were definitely worse than I expected.

PREDICTION-0021: 2020-06-30: Multi-core performance of the first Apple Mac ARM chips

  • From: @ebruchez
  • Text: “The first ARM-based Macs released will include a machine with a multi-core Geekbench score of at least 10000.”
  • Comments:
    • The fastest MacBook Pro multi-core score is 6958 for the “MacBook Pro (16-inch Late 2019) Intel Core i9-9980HK @ 2.4 GHz (8 cores)”.
    • The fastest Mac multi-core score if 19155 for the “Mac Pro (Late 2019) Intel Xeon W-3275M @ 2.5 GHz (28 cores)”.
    • The fastest multi-core score is 23944 for the “AMD Ryzen Threadripper 3990X 2.9 GHz (64 cores)”.
  • Resolution:
    • FAIL: I was further away with this one than for the single-core performance. The Mac Benchmarks shows 7396 (while I predicted 10,000) for the Mac Mini.
    • UPDATE: 2021-10-20: The M1 Max appears to have a multi-core Geekbench score of 12693. This clearly beats my prediction, 16 months after the introduction of the M1.

PREDICTION-0020: 2020-06-30: Single-core performance of the first Apple Mac ARM chips

  • From: @ebruchez
  • Text: “The first ARM-based Macs released will include a machine with a single-core Geekbench score of at least 1840.”
  • Comments:
    • Apple announced their first ARM-based Macs for the end of 2020.
    • The current rumors are for the first machine to be the 13” MacBook Pro, followed by an iMac.
    • Current scores for reference
      • The fastest MacBook Pro single-core score is 1227 for the “MacBook Pro (13-inch Mid 2020) Intel Core i7-1068NG7 @ 2.3 GHz (4 cores)”. So a 50% faster single-core score would be about 1840.
      • The fastest ARM CPU single-core score is 1327 for the “iPhone 11 Pro Apple A13 Bionic @ 2.7 GHz”. This means that Apple would have to increase their single-core by 40% between this CPU and their new CPU.
      • The fastest single-core score overall is 1419 for the “Intel Core i9-10900K 3.7 GHz (10 cores)”. (This means that Apple would have to increase their single-core by 30% between this CPU and their new CPU.)
  • Resolution:
    • FAIL: I was pretty close with this one! As of 2021-02-06, Geekbench shows some 1744 scores (while I predicted 1840). The Mac Benchmarks shows, however, 1708 for the Mac Mini, a little lower.
    • UPDATE: 2021-10-20: The M1 Max appears to have a single-core Geekbench score of 1783. This doesn’t beat my prediction, 16 months after the introduction of the M1.

PREDICTION-0022: 2020-07-09: Donald Trump will lose the 2020 US presidential election

  • From: @ebruchez
  • Text: “The November 2020 US presidential election will take place in 2020. Incumbent Donald Trump will lose to his opponent.”
  • Comments:
    • As of 2020-07-09
      • FiveThirtyEight shows Donald Trump low in popularity (~60% disapproval).
      • FiveThirtyEight also shows Donald Trump way behind in national voting intentions.
      • Joe Biden is the presumptive Democratic nominee.
    • In the 4 months to the election, a lot can change. But it doesn’t look good for Trump as of July 2020 and I am predicting (and also hoping for) his loss.
    • I think that polls indicate a very different climate compared to 2016. Joe Biden is not Hillary Clinton. The pandemic and other issues have been very badly managed by Trump and to a point where these issues are very visible to voters. I only see more reasons for voters to vote against him and for his opponent than in 2016.
    • While I can understand voters who wanted change in 2016 and try an outsider (no matter how flawed), more than 3 years in I believe that the proof is in the pudding: we have seen such a litany of incompetence, corruption, bad ideas, bad execution, lack of leadership, lack of vision, and general moral bankruptcy, that I think that a significant number of 2016 Trump voters have changed their mind at this point.
    • Keeping in mind that Trump won fairly narrowly in 2016, I do think that there will be a swing in opinion significant enough to make him lose this time.
  • Resolution:
    • SUCCESS: 2020-11-07: All major US TV networks have called the race in favor of Joe Biden. It took days as lots of mail ballots were cast, and the election was pretty close in many states. But in spite of recounts and litigation, it appears now impossible to see a reversal as Biden has won too many states and is now way beyond the 270 electoral votes needed.
  • Thoughts:
    • The final result of the election could have gone either way, in the sense that so much is unpredictable in life and politics. However, my general gist was correct: people did turn out in massive numbers for what amounts to rejecting Trump. One could have hoped for a more massive rejection, of course, but I, for one, will take what I can.

PREDICTION-0023: 2020-10-27: 2020 Presidential election

  • From: @avernet
  • Text: “The election will be mostly fine. It’s going to be like Y2K: a lot of worries but in the end no big problem. Maybe there will be a little violence and incidents but that’s all.”
  • Resolution:
    • SUCCESS: 2020-11-07: There was no major incident on election day and during the days after.

PREDICTION-0002: 2013-09-16: iPhone CPU performance

  • From: @avernet
  • Text: “No 40x CPU perf improvement between iPhone 5S to 6 years after its release.”
  • Comments:
    • 2007-06-29: Original iPhone release.
    • 2013-09-20: iPhone 5S release (just over 6 years later).
    • On 2013-09-10, Apple announced the iPhone 5S and and claims “CPU is 40x faster than original iPhone, with half coming with jump between A6 and A7”.
    • Geekbench: 11.8x (1619 for the iPhone 5, 137 for the original iPhone)
    • NOTE: This is not about overall performance, although that is interesting, but about CPU performance.
    • It’s almost certain that there will be devices called “iPhone” in 6 years.
    • The question is: how do you measure performance improvements? On a per-core basis seems unfair. So some kind of reasonable benchmark, possibly using all the cores, should show a 40x performance improvement.
  • Resolution:
      • The iPhone 11 Pro was released on 2019-09-20, exactly 6 years after the iPhone 5S. There was nothing near 40x improvement in CPU performance, although the improvements have been significant:
        • single core
          • iPhone 5S: 259
          • iPhone 11 Pro: 1327 (or a 5x factor)
        • multi-core
          • iPhone 5S: 476
          • iPhone 11 Pro: 3385 (or a 7x factor)
    • HOWEVER, the Metal performance has seen a huge improvement:
      • iPhone 5S: 20
      • iPhone 11 Pro: 6248 (or a 312x factor)

PREDICTION-0004: 2013-12-03: Amazon Prime Air timeline

  • From: @avernet
  • Text: “Amazon Prime Air won’t be usable in 5 years at Thomas’s place (for technological or cost reasons, including but not limited to FAA requirements or high initial cost).”
  • Resolution:
    • SUCCESS: There is definitely no such service available in the Bay Area here now.
    • “Amazon Prime Air is also the arm of the company working on their the drone-based delivery system—which has long been in development after its initial announcement in 2013, but has essentially stalled due to U.S. airspace regulations.” Observer article from 2017-11

PREDICTION-0013: 2015-12-03: Swift in the browser

  • From: @ebruchez
  • Text: “Now that Swift is actually open source, developers, whether at Apple or third-party, will produce a usable and efficient Swift-to-JavaScript compiler similar to Scala.js before the end of 2016. It will be called Swift.js.”
  • Comments:
    • I will consider this prediction successful even if the compiler is not officially called Swift.js!
    • The intent of this prediction is that Swift.js will reuse much of the open source Swift toolchain, in the same way that Scala.js reuses much of the Scala toolchain. (As a counter example, there is at least one project attempting to do this on GitHub, but it implements a subset of Swift and appears very immature so doesn’t qualify.)
    • Swift already has the Swift Intermediate Language which could be the basis for a JavaScript backend.
    • Swift.js as described here doesn’t compile to asm.js/WebAssembly, but maps to JavaScript in a way very similar to what Scala.js does.
    • One important difference is that Swift.js will be garbage-collected, while its LLVM version is reference-counted. Developers will have to be made aware of that difference.
  • Resolution:
    • FAIL
    • 2017-01-21: Nothing has come from Apple on this front, and nothing serious has been happening outside of Apple either. There is ShiftJS but it is in “early development” and it doesn’t appear to reuse any of the existing Swift compiler. So it doesn’t qualify.
    • I still think tha this will happen, but it is clearly lower priority for Swift compared to Swift on the server.

PREDICTION-0010: 2015-06-11: A Node.js for Swift

  • From: @ebruchez
  • Text: “Now that Apple has announced the open sourcing of Swift, I predict that by the end of 2016 there will at least one popular server-side, async/event-driven platform a la Node.js but based on Swift. It will run on Linux and OS X.”
  • Updates
    • 2016-03-01: Swift Express, a Play- and Express.js-inspired framework.
  • Resolution:
    • 2017-01-20: There are a number of server-side Swift frameworks (see also The Awesome Server Side Swift List):
      • Perfect, on GitHub
        • 10,300+ stars on GitHub
        • Apache license
        • async
      • Vapor
        • 8,200+ stars on GitHub
      • Kitura
        • by IBM
        • 5,200+ start on GitHub
      • Now how to judge “popular”? For reference, Node.js has about 30K stars on GitHub. I think these frameworks can qualify as popular and Node.js-like, although I realize that don’t have the impact that Node.js or Ruby on Rails have had, and far from it.
      • Chris Lattner in this interview confirms that server-side Swift was/is the next big step towards Swift’s total world domination goal.
      • It could be (my opinion) that Apple is working on a major server-side platform which will dwarf what we have seen so far.
      • I am claiming this prediction as a success, while realizing that there is still a lot of future potential for growth for Swift-based server-side platforms.

PREDICTION-0009: 2015-01-21: Marijuana in California

  • From: @ebruchez
  • Text: “California will legalize marijuana before year 2020.”
  • Resolution:
    • 2017-01-21: Well in advance of 2020, Californians have approved legalization of recreational marijuana on 2016-11-09.

PREDICTION-0014: 2016-08-02: 2016 election will not be close

  • From: @avernet
  • Text: “The 2016 US presidential election will not be close: Trump will lose by a significant margin.”
  • Resolution:
    • FAIL
    • 2017-01-21: Needless to say, this didn’t happen.

PREDICTION-0019: 2016-09-02: general-purpose jack adapter for the iPhone 7

  • From @avernet
  • Text: “There is no way the iPhone 7 will ship with a general-purpose Lightning ↔ jack adapter for corded headphones.”
  • Resolution:

PREDICTION-0012: 2015-07-04: Swift overtaking Objective-C

  • From: @ebruchez
  • Text: “In June 2016, two years after Apple’s announcement, Swift will have overtaken Objective-C on GitHub based on metrics by RedMonk.”
  • Comments:
    • This is a renewal of PREDICTION-0005 below set for one year later.
    • I think the momentum of Swift will accelerate this year due to Swift 2, open sourcing, and the fact that a rise in Swift means a decline of Objective-C.
    • Last year’s PREDICTION-0005 was too aggressive. I still think the updated prediction might not be realized in one year, but this is my bet!
  • Resolution:
    • FAIL
    • RedMonk: “Swift at this point has become the canonical example for the inertia of incumbent languages. Have followed an unprecendented growth trajectory since its introduction, this run is the first in which Swift has not gained but merely held its position of #17. […] But it’s clear that further gains for Swift will not come easily, and will instead be the product of widespread usage across an array of communities. As discussed in the last iteration of these rankings, Swift has opened up new avenues for growth beyond iOS development via its release as open source software and the embrace of third parties like IBM or Perfect, but these have yet to yield gains in new discussion or code sufficient to propel it forward in these rankings. We’ll be watching for signs of this type of new growth closely.”
    • Ok, so I was definitely too aggressive again, and Objective-C still leads. I am sure Swift will overtake Objective-C within a few years, but I cannot guess where to put the line now. Another 2 years? 5 years? Longer?

PREDICTION-0011: 2015-07-02: Timeline of real-life test of Google self-driving cars

  • From: @avernet
  • Text: “Google will not manage to start a real-life test of their self-driving car in the US before the end of Q1 2016. By real-life, we mean that such a test would not rely on Google employees. The test doesn’t have to be in California, but it must be in the US and on public roads.”
  • Comments:
    • This is in response to a rumor that Google might start such a test in 2015.
  • Resolution:
    • 2016-08-03: As far as we can tell, Google hasn’t done such a real-life test by 2016-03-31.

PREDICTION-0008: 2015-01-09: iPhone with 2 GB of RAM

  • From: @ebruchez
  • Text: “At least one new fall 2015 iPhone model will have 2 GB of RAM or more.”
  • Resolution:
    • Both the iPhone 6S and iPhone 6S Plus have 2 GB of RAM.

PREDICTION-0006: 2014-09-30: Share of smartphones with large screen sizes

  • From: @ebruchez
  • Text: “In 1 year, 50% of smartphones sold in the US in the previous quarter will have screen sizes around 5.5 to 6 inches.”
  • Comments:
    • In other words, I am predicting that the larger form factors will gain a lot of traction quickly.
    • “Smartphone” means the device has a connection to a traditional cellular network (3G, LTE).
  • Resolution:
    • FAIL
    • According to this, on the Android side, the share of “phablets” is only 13.7% as of June 2015, although it doubled since early 2014.

PREDICTION-0005: 2014-06-11: Swift overtaking Objective-C

  • From: @ebruchez, @avernet
  • Text: “In June 2015, one year after Apple’s announcement, Swift will have overtaken Objective-C on GitHub based on metrics by either of RedMonk or Ohloh.”
  • Comments:
  • Updates:
    • 2015-01-16: “Swift just jumped from number 68 to number 22 in […] RedMonk” (Wired, RedMonk). Ojbective-C is still in position 10 though, so I don’t know if the prediction will come true.
  • Resolution:
    • FAIL
    • The 2015-06 RedMonk survey places Swift at position 18, and Objective-C still at position 10, and GitHub reflects this. So my prediction has obviously failed. But RedMonk says of Swift:

      “Even if you assign little importance to the actual ranking, then, there is no debate that Swift is growing faster than anything else we track. The forthcoming release of Swift as open source and availability of builds for Linux, as well, should theoretically provide even more momentum going forward.”

    • One interesting aspect of the respective ranking of Objective-C/Swift is that the languages are linked and the rise of Swift will happen at the detriment of Objective-C. So as Swift becomes more popular, Objective-C will become less so.
    • I think the momentum of Swift 2 will be big, causing many iOS programmers to consider a switch to Swift. I am renewing my prediction for next year as PREDICTION-0012.

PREDICTION-0007: 2014-12-29: Apple open sourcing Swift

  • From: @ebruchez
  • Text: “During 2015, Apple will either open source or unequivocally announce that it will open source Swift.”
  • Resolution:
    • 2015-06-08: Correct prediction as announced by Apple at WWDC 2015!
    • 2015-12-03: Open-sourcing is effective on GitHub.


2015-07-04: General considerations on self-driving vehicles

Self-driving cars are a great area for predictions!

The technology of the Google driverless car is pretty impressive already, based on this June 2015 TED Talk by Chris Urmson. I am pretty sure that self-driving cars will happen, although some legal, social, and technological obstacles remain. The big question is when they will be overcome, and what kind of predictions make sense.

Here are some examples of milestones that come to mind:

  • Demonstration of end-to-end automated vehicles picking up and dropping off people in realistic scenarios (such as home to work). This must include the ability for the vehicle to park or wait in a safe location, and to drive safely and without unreasonable delays. This must include local driving, and may include highway driving. For demonstration purposes, a human supervisor may be present in the car.
  • First US state to allow (under conditions to be met) certain types of fully automated vehicles on at least some public roads. Such vehicles do not require a driver at all, and might not even have a driving wheel (like the Google car). No driver license is needed to enter the car. The car can circulate on its own without any human aboard.
  • First lawsuit related to to a self-driving vehicle having allegedly caused harm.
  • First company operating a fleet of fully automated taxis (it could be Uber).
  • Self-driving vehicles (owned or not) are an option for some people (in the same way electric vehicles such as the Nissan Leaf are as of 2015) as a replacement for a regular car. Such vehicles might be operated by a separate company. They are an option to take people to and from work on a daily basis, as well as perform other trips.
  • Self-driving vehicles are mainstream.
  • Rate of car ownership in the US is reduced by 50% compared to 2015 levels.
  • Number of cars on the road in the US is reduced by 50% compared to 2015 levels.
  • Merger/bankruptcy of large car manufacturers due to the reduced market for vehicles.
  • Cities start reclaiming and repurposing roads and unused parking lots.
  • Most (more than 50%) people don’t own a car.
  • It is reasonable to most young people (at least in a certain area) to completely avoid getting a driver license and never learn how to drive.
  • Some public roads are closed to vehicles with human drivers.

Trucks are also an obvious target for self-driving technology as they drive long distances. Here are a few milestones:

  • Demonstration of self-driving 18-wheeler for long highway drives (such as CA highway 5). Daimler has just revealed a limited concept in May 2015. See also Self-Driving Trucks Are Going to Hit Us Like a Human-Driven Truck.
  • Commercial availability of self-driving 18-wheeler for long highway drives (such as CA highway 5), still requiring a driver in the cab for supervision, entering and existing highways, and local driving.
  • First law against self-driving trucks passed in a US state, under the pressure of truckers.
  • Repeal of such laws after legal battles.
  • Fully automated self-driving 18-wheeler able to do end-to-end trips.