Recently, we did an experiment: We took an outdated issue of a respected popular science magazine, Scientific American, and researched exactly what happened to the highly-touted breakthroughs of the era that would supposedly change everything. What we discovered is just how terrible we are at predicting the long arc of scientific discovery.

The daily churn of science news tends toward optimism. You know what I’m talking about: New cure! New breakthrough smashing Moore’s law! New revolutionary technology! I write about science, and I am always uncomfortable trying to predict how a new piece of research will change the future.

That’s because science can be wrong. It can go down dead ends. And even when it doesn’t, almost everything is more complicated and takes longer than we initially think. But just how wrong and how long?


We can’t very well time travel to the future for those answers, but we can look backward. I recently dig up the 2005 December issue of Scientific American and went entry by entry through the Scientific American 50, a list of the most important trends in science that year. I chose 2005 because 10 years seemed recent enough for continuity between scientific questions then and now but also long enough ago for actual progress. More importantly, I chose Scientific American because the magazine publishes sober assessments of science, often by scientists themselves. (Read: It can be a little boring, but it’s generally accurate.) But I also trusted it not to pick obviously frivolous and clickbaity things.

Number one on the list was a stem cell breakthrough that turned out to be one of the biggest cases of scientific fraud ever. (To be fair, it fooled everyone.) But the list held other unfulfilled promises, too: companies now defunct, an FBI raid, and many, many technologies simply still on the verge of finally making it a decade later. By my count, only two of its 16 medical discoveries of 2005 have resulted in a drug or hospital procedures so far. The rosy future is not yet here.

Science is a not a linear march forward, as headlines seem to imply. Science is a long slow slog, and often a twisty one at that. That’s obvious in retrospect, when we can see the dead ends and the roadblocks. It’s less obvious looking ahead, as we’re being bombarded with promising new drugs and wondermaterial breakthroughs. So let’s take a look together.


The December 2005 Scientific American 50 lists winners—people, organizations, and companies—spread out among 17 trends. In my rundown below, I’ve omitted a few of the winners that are explicitly about policy and business (the Kavli Foundation, NGOs in China, etc.) to focus on the science and technology. In some cases, I’ve also condensed people or companies that had made similar findings into one entry.

Stem Cells

Prediction: Patient-specific stem cells that pave the way for stem cell therapy

In May 2005, South Korean researcher Woo Suk Hwang stunned the world by making stem cells that perfectly matched the genetic profiles of 11 patients—the big first step toward using cloning to treat diseases. Just few months later, he would stun the world again when it turned out he had faked his remarkable results.

Hwang was discredited and the cloning technique he championed quietly slipped out of the scientific mainstream. That’s not because it doesn’t work—researchers have since shown it can.
Hwang’s fraud actually obscures a subtler and more mundane fact about how promising scientific research hits dead ends.


There are now two ways to make stem cells, you see. The cloning technique used by Hwang involves transferring the nucleus of a patient’s cell into an embryonic stem cell. This requires destroying human embryos—the crux of the stem cell controversies of the early 2000s. In 2006, though, researchers found that mature cells can also be reprogrammed it into stem cell, creating so-called induced pluripotent stem (iPS) cells. The ethical controversy over stem cells evaporated. In September 2014, a blind Japanese women became the first patient to receive iPS cells as treatment. Therapeutic cloning ultimately may not be a dead end, but the bulk of funding and research is now focused on iPS cells. The newer technique is both less ethically and technically fraught. But back in 2005, cloning was shining so brightly as a star.

Flu Vaccines

Prediction: A low-dose flu vaccine


An intradermal flu vaccine was approved in 2011. They’re still not common, but some pharmacies and doctor’s offices offer them because they require a smaller needle. The smaller dose could also come in handy during outbreaks when supplies are low.

Prediction: Vaccine patches

IDEO’s design for a vaccine patch. Back when the technology was still promising, the company behind the patch, Iomai, was acquired by Intercell. Then the patches tanked in clinical trails. Intercell was then itself acquired for less money than it bought Iomai. IDEO


Vaccine patches are no longer a thing after a series of failed clinical trails. Back in 2005, Iomai Corporation’s bright idea was to supplement a low-dose vaccine with a patch containing adjuvants, molecules that enhance immune response. Despite a promising initial trials and a big government contract, clinical trials ultimately found the flu vaccine patch did not work. A similar patch for a traveler’s diarrhea vaccine also failed in 2010. The company that acquired Iomai for $189 million was later itself acquired—for less money.

Prediction: Anti-flu drug that works on multiple viruses


Fludase is a prospective drug that works against multiple flu viruses by targeting the lung cells rather than viruses. Ten years later, it is still in Phase II trials. But the company behind it, NexBio, Inc., has had other things to worry about it. The FBI raided its offices in 2011 and fined the company $2 million fine overbilling the government on Fludase research.

Solar Cells

Prediction: Solar panels that absorb infrared light


Still working on it.

Prediction: More efficient dye-based solar cells

Still working on it.


Prediction: Hybrid solar cell that generates and stores electricity

Still working on it.


Prediction: Storing solar energy in hydrogen fuel

Hydrogen Solar, the British company that announced a 10-fold improvement in making hydrogen fuel back in 2005, has been renamed Stored Solar, which seems now to exist only as a shell of a website. The idea of solar energy hydrogen fuel is still around, though, and guess what? They’re still working on it.


We should note, though, that while these new ways of making solar cells have not yet made it to market, solar energy hasn’t been a total dud. Huge commercial solar power plants have come online recently. The cost of existing solar panel technology has gone down drastically, and creative financing solutions pioneered by SolarCity had made residential solar a lot more appealing.

Repairing the Heart

Prediction: A chemical cocktail that gets heart cells to divide

Still working on it. A 2005 study identified a particular growth factor that could get rat heart cells to divide in a dish. Getting it to work in humans and inside a live human heart that has to keep beating seamlessly has been a lot more challenging.

Flexible Electronics

Prediction: Plastic semiconductors that make for flexible electronics.



Still working on it. Silicon is brittle, so for flexible electronics, you need semiconductors made from carbon-based polymers. Polymer Vision and Plastic Logic became the first two companies to commercialize the technology, launching bendable e-readers in 2007. Both have since abandoned the e-readers, though plastic electronics companies are now jumping on the wearables trend.

HIV Treatments

Prediction: HIV infections begin in the gut.

This finding hasn’t yielded breakthroughs in preventing or eradicating HIV infections yet, but the finding dovetails nicely with the recent rise of the gut microbiome in medicine. It turns out HIV-infected people have very different gut microbiomes than health individuals.


Prediction: New drug target called DHS

Still working on it. A paper was published in 2014 investigating new compounds to target DHS after previous ones failed.

Wireless Technology

Prediction: City-wide “super Wi-Fi” to take over old TV spectrum



With analog TV was going bye-bye, the FCC in 2002 suggested keeping one portion of the unused TV spectrum open for “super Wi-Fi” that can cover entire cities rather than just a houses. So far, this super Wi-Fi has only been deployed in two test cases: Wilmington, NC beginning in 2012 and West Virginia University beginning in 2013.

Prediction: Fractal antennas in cars

The shark fin antennas that you sometimes see on cars? Those are fractal antennas, which are compact devices that can receive at multiple frequencies. In 2005, manufacturers started putting them in cars. While still not universal, fractal antennas have become more common with the proliferation of networks for a car to connect to: radio, GPS, Bluetooth, and so on.

Artificial Life

Prediction: “Plug and play” genetic circuits


A decade ago, scientists created a genetic switch to toggle the production of a protein on or off. Now, these circuits have gotten more sophisticated, and they can act as various types of oscillators. The idea is that you can plug many of these circuits together are get, say, a smart plant that acts a pH sensor. However, scientists have also run into problems when a gene circuit put inside a new organism does not work as expected.

Prediction: Synthesize long strands of artificial DNA


Scientists have gone from synthesizing several genes worth of artificial DNA in 2004 to synthesizing an entire artificial genome in 2010.

Prediction: Laboratory-synthesized cell membranes


Still working on it.

New Aircraft

Prediction: Airbus A380, the world’s largest passenger airliner ever

Despite being ultraquiet and light, the Airbus A380 has been well, a “commercial disaster.” Airbus made the wrong bet with a massive plane designed to shuffle people between hubs. It turns out passengers actually prefer direct flights on smaller planes.


Prediction: A plane that runs on bi0ethanol alone

Brazil’s EMB 202 Ipanema crop duster was one of the first planes to run on ethanol alone. Other engines have since been modified to run on ethanol, too, but ethanol-fueled planes have not and probably will not catch on in the U.S. for a simple reason: Brazil is in a unique position because its plentiful sugarcane can be efficiently converted to ethanol fuel—while corn, plentiful in the U.S., cannot. That’s unless there’s a big breakthrough in second generation bioethanol.

Brain Scanning

Prediction: Single-neuron functional imaging inside a live rat brain.



Neuroscientists have since improved their techniques to pull off whole-brain functional imaging of every single cell in the zebrafish brain. However, zebrafish brains are considerably smaller and easier to image than those of rats.

Prediction: Using fMRI brain activity to decode what a person is looking at

In 2005, researchers were able to use functional magnetic resonance imaging of the brain’s visual cortex to figure out which one of 8 visual patterns a person was looking at. In a more recent 2011 paper, researchers were even made to decode movies from brain activity.


Prediction: Tracking “amyloid plaques,” a hallmark of Alzheimer’s, in the brains of live mice

A version of this technique is now used to help diagnosis Alzheimer’s humans. Even though the technique works, the test is not perfect: The presence of amyloid plaques also does not perfectly correlated with cognitive impairment decline.


These three brain tracking advancements were all significant, but this list is missing what is perhaps the most important neuroscience paper that year, even the decade—a breakthrough that would let neuroscientists go from merely imaging the brain to precisely controlling it. That would be a technical report in Nature Neuroscience that describes using algal proteins to make neurons light sensitive. It would go on to birth optogenetics, a technique that has since become widely used in everything from implanting false memories to turning off pain in mice. The most revolutionary ideas can be hard to miss.

Carbon Nanotubes

Prediction: New ways to make and process carbon nanotubes for commercial electronics


Thin sheets of carbon atoms rolled up into a cylinder—aka carbon nanotubes—have lots of unique electrical and mechanic materials. But this wonder material of the early 2000s hasn’t really caught on. The applications touted in SciAm—a Motorola television screen, flexible electronics, nanotube arrays, and nanoscale circuits—have not made it into your living room. Scientists have since found reliability problems with nanotubes in electronics, but the ultimate problem may just be lack of infrastructure for bringing the technology to market.

There’s another undercurrent running through the past decade, which is that graphene has stolen a lot of carbon nanotubes’ buzz as a wonder material. Though now, graphene is also running headlong into the challenges of commercialization.

Gene Therapy

Prediction: Discovery of genes for deafness.


In 2005, a couple of papers were published each showing how a particular gene controlling the growth of inner hair cells could be manipulated to treat deafness. A therapy based on the more promising of the two, a gene called Atoh1, entered clinical trials in humans at the end of last year. The second, called, Rb1, is still being studied in rats.

Prediction: Using silica particles for gene therapy


Still working on it.

Silicon Lasers

Prediction: Silicon lasers that could lead to high-speed chips.


Still working on it. Nine years later in 2013, researchers finally made a micrometer-sized version of a silicon laser, which would actually be small enough to put inside a device. Intel has a Silicon Photonics Solutions Group dedicated to bringing the technology to market.