Predictions For 2021

2017 article, not via Zoom

Allan Lichtman correctly predicted the 2020 presidential election, based on a 7-6 edge in “keys” to Joe Biden. His model, which he developed in 1981 with the late Vladimir Keilis-Borok, simply says that the incumbent party will lose the presidential election if 6 or more keys go against them. It has been right in every election since 1984.

Today we make some predictions for the new year, 2021.

Lichtman has not ventured a prediction for today’s Senate runoff elections in Georgia. Last November 19, he penned an article eviscerating the performance of poll aggregation in the November 3 election. We have considered writing a post about that and about statistical predictions during the pandemic, in the direction of taking more moderate assessments of how various polls, predictions, and policies have fared.

Polls of today’s races currently show a 1–2 percentage point edge for the Democratic candidates, but there are multiple caveats: The difference is just about one standard deviation, well within the margin of error even after aggregation. The polls have gyrated with multiple lead flips this past month; two polls from the past few days collected here show both Republican candidates tied or ahead. Many pollsters have avoided these races, and the five most recent polls all have sample sizes well under 1,000. The count beginning tonight may show a zigzag blue-to-red-to-“blur” shift for reasons of which votes are tallied when that we discussed generally here.

With that said, let us address non-election predictions.

Previous Predictions

Back in 2019 we made some predictions that were easy:

• No circuit lower bound or better will be proved for SAT. Well that’s a freebie.

• A computer scientist will win a Nobel Prize. No—indeed, less close than other years.

• At least five claims that and five that will be made. Gerhard Woeginger’s claims page is still paused at 2016, so we can only judge from our own window of communications made to us or commented in the blog. All we can say is that the number in each case is “order-of” .

• A “provably” secure crypto-system will be broken. For this one we don’t have to check any claims. We just pocket the “yes” answer.

We skipped 2020—maybe that was smart of us? Our point about predictions in 2019 was the importance of “when” as opposed to “what.” With regard to some predictions, we can take the pandemic is as a global postponement of “when.”

We claim complete and ongoing timeliness with another 2019 prediction, however. In full, we wrote:

• Quantum supremacy will be proved—finally. But be careful: there is a problem with this whole direction. See the next section.

The section two years ago talked about the difficulties of proving such a claim. A claim was duly made in October 2019, which we evaluated here, but arguments over “proof” have not abated. They have just recently been summarized by Gil Kalai here, and also here by Scott Aaronson in regard to an independent claim by researchers from USTC in China.

Predictions for 2021

Here are some new predictions for 2021. We will try and skip easy ones like the above. We will try to make some that are surprising. We’ve made predictions about evidence for life on exoplanets before; we do something different there too. We do not venture any about the course of the pandemic.

1. Quantum algorithms: will be proved to be in sub-exponential time.

2. Further work with AlphaFold will lead to a proof of a sub-exponential time algorithm for protein folding.

3. In consequence or independently, a SAT algorithm will be discovered that runs in sub-exponential time, thereby refuting various forms of the Exponential Time Hypothesis. What do we mean by “sub-exponential” time here? Dick says the time for the SAT algorithm will be . Ken more modestly says time .

4. Another Clay prize problem (besides versus ) will be claimed as solved, and then retracted.

5. A new kind of sorting network of size will be discovered, inspired by a connection to integer multiplication.

6. A new factoring algorithm will be discovered that beats the number field sieve.

7. A lower bound argument will show that clique requires formula size over the basis.

8. Significant chemical evidence will be observed for the existence of life on a non–exoplanet.

9. Dogs will reach 25% of the proficiency of cats at navigating home obstacle courses, but not before the third quarter of 2021.

10. Privacy-preserving apps for doodling during Zoom meetings will become hot sellers by midyear.

Open Problems

What are your predictions for 2021?