The Evolution of Bacteria & Outcome-Based EdTech Firms
I am going to start by (literally copying) a story from Mr. Pulak Prasad’s book ‘What I Learned About Investing from Darwin’
A scientific experiment that had been going on since 1988
Tim Cooper had no idea that that cold and windy Saturday morning in a lab at Michigan State University in January 2003 would be one of the most important of his life.
Tim started performing his well-rehearsed routine. He had done it dozens of times over the past three years, but he knew that he had to be very careful. He was. The experiment had been going on for fourteen straight years, and he was not going to be the one responsible for any mishap.
First, he took a set of twelve new flasks and carefully measured exactly 9.9 milliliters of fluid into each. Next, he went to the incubator and removed the twelve old flasks housing generation number 33,127. He would inoculate the new flasks with 0.1 milliliters of fluid from each of these twelve old flasks. But he needed to check the old flasks first. He picked up two of them and saw what he expected. The next two flasks also seemed acceptable. But in the third set of two flasks, in the flask labeled “Ara-3,” he saw that the fluid had turned opaque instead of being mildly cloudy like in the other old flasks. That shouldn’t have happened. The lab had seen similar problems in the past owing to contamination. There was a strict protocol for solving the issue, and Tim was well-versed in it. Tim replaced the “faulty” old Ara-3 flask and came back on Sunday to check the outcome. Of course, he expected the usual result.
But he was in for the same surprise: The new Ara-3 flask, too, had turned turbid. Something was very wrong. Or very right.
Lessons from Lenski
On February 24, 1988, Professor Richard Lenski seeded twelve sterile flasks containing ten milliliters of glucose solution with the common bacterium Escherichia coli (commonly called E. coli).
Lenski started an experiment that continues today and has yielded hundreds of research articles, dozens of doctorates, and global acclaim.
He started an experiment on long-term evolution with two key differences. First, he wanted to use microscopic organisms that have a generation time of only twenty minutes, thereby allowing tens of thousands of generations to evolve over a human lifetime. Second, unlike the artificial selection experiments on foxes (or dogs or wheat), in which the experimenter chooses who (or what) reproduces, Lenski wanted the experimental environment to choose.
When the experiment started, all twelve flasks had genetically the same E. coli since all cells came from the same mother cell. Each flask contained hundreds of millions of bacteria, so there was ample opportunity for mutations to emerge. An essential experimental twist was that food was in limited supply. Every day, the E. coli population increased for about six hours until the glucose was exhausted. At this point, the bacteria would stop dividing and wait. The next day, a lab member — someone like Tim Cooper — would take 0.1 milliliters out of each flask (1 percent of the flask’s content) and inoculate a new flask containing 9.9 milliliters of fresh glucose solution. A new cycle would then begin until the next day when the same procedure would be repeated. And the next day. Week after week, month after month, year after year.
What he found was remarkable. In 2011, after fifty thousand generations of evolution, he said, “To my surprise, evolution was pretty repeatable. . . .Although the lineages certainly diverged in many details, I was struck by the parallel trajectories of their evolution, with similar changes in so many phenotypic traits and even gene sequences we examined.”
It wasn’t that the bacteria had not evolved. On the contrary, they had evolved a lot. The E. coli in different flasks adapted differently to the starvation diet by either growing faster or becoming more numerous than earlier generations. But the general trend was unmistakable: On average, the population grew 70 percent faster than their founding ancestors. In addition, the researchers found another example of convergent evolution. All twelve populations of bacteria had lost the ability to synthesize a sugar called D-ribose because all had undergone the same set of genetic changes.
One day, fifteen years after the start of the experiment, the reason the Ara-3 flask had turned opaque was that, unlike the other flasks, it had experienced a population explosion. The Ara-3 population was ten times the size of the populations in the other flasks.
How could this have happened when all flasks had only a limited amount of glucose?
The Ara-3 bacteria had developed the ability to feed on some other ingredient in the solution.
The only other candidate was a molecule called citrate present in all glucose solutions from day one. However, at the time of the beginning of the experiment, E. coli was explicitly known not to synthesize citrate in the presence of oxygen. Therefore, this inability was used to identify whether a type of bacteria was E. coli or not! The 33,127th generation of Lenski’s bacteria had deviated drastically from the clean and simple story of inevitable convergence. It was as big a case of divergence as biology had ever seen experimentally. Lenski’s lab later found that the ability of the Ara-3 bacteria to digest citrate resulted from a series of mutations after about twenty thousand generations. Unfortunately, each mutation was rare, so no bacteria in any of the other flasks developed this ability. It is worth paying attention to the two critical lessons from Lenski’s long-term experiment: 1. Convergence is the dominant pattern in the natural world. 2. On rare occasions, it isn’t.
Prasad, Pulak. What I Learned About Investing from Darwin (p. 259). Columbia University Press. Please buy the book, it is amazing!
The Promise of Outcomes in Ed-tech
Several companies across the globe started with the promise of driving outcomes in edtech.
Surprisingly almost all the companies underwent similar changes irrespective of the course offering!
- Everyone realized that everyone can not be guaranteed outcomes. It is several interviews that can be assured. So most companies promise several interviews or shortlists and not the job anymore.
- All the companies realized that to drive outcomes, real work needs to happen so they have features like mentors, mock interview practice with a real expert, real feedback, proper resume making, dedicated counselors (usually 1 to 10 to 40 mapping), etc. So almost all the outcome-focused companies have designed a very authentic learning experience that is 2x to 10x better than comparative degree programs at Great Learnings or upGrad.
- To drive outcomes, candidates need to put in the effort. So almost every company has introduced certain criteria that learners need to meet such as progress, consistency, attendance, and performance in certain tests and quizzes to become eligible for outcome. And most candidates fail to meet these criteria.
- If you interview 20 candidates from any of these companies, you will hear ditto same responses i.e. learning was great, mentors were great but the outcome was not up to the mark.
- Session attendance is a major problem for all of them esp. some candidates have a day job.
All outcome-based companies are evolving exactly in the same manner irrespective of whether they teach software engineering, analytics, product management, or digital marketing!
And this is irrespective of the model they operate from i.e. relevel, upraised, newton school, analytics vidya, able jobs ==> All of them have different models e.g.
- Relevel ==> offered free placement services (5 interview shortlist) if you crack their exam. Relevel makes money if you are not qualified for the test and opt for the course.
- Upraised ==> offered a free test that helps you assess your skills. If you are qualified then only you could pay and join the course! (Big contrast to Relevel where the course is exclusively for people who don’t qualify).
- Newton School==> offers admission to everyone but there are performance criteria for you to become eligible for outcomes. Anyone and everyone can join by paying the fees but they can only get outcomes if they perform during the program.
- Simplilearn ==> Offers admission based on a generalist aptitude test and there are performance criteria. The courses are usually low on personalization and the company usually follows the process of intaking people and processing refunds quickly; they have kept variable costs (usually associated with personalization) on the lower side.
This phenomenon is not unique to this case. It happens across all industries, all stages of the companies, almost all the time.
Drawing Analogy with Lenski’s Experiment
Bacteria under similar environments (structural circumstances & constraints) evolved similarly despite having many variations. In the same ways, edtech companies focused on outcomes are evolving in a remarkably similar manner!
Convergence is the dominant pattern in the natural world but on rare occasions, it isn’t.
As variations continue, some businesses may keep on trying different pivots. It may result in the creation of something amazing which can factor in the limitations of today and can survive and thrive!
We all will see how it goes.
