According to the American Heritage Dictionary, the gatekeeper is one who is in charge of a passage through a gate, one who monitors or oversees the actions of others. Research from the 1940s portrayed housewives as gatekeepers in charge of the family food channel during World War II. The technological gatekeeper, a couple decades later, characterized individuals participating in the flow of scientific and technological information through organizations.Gatekeepers kept on reappearing with their different clothing and methods, as a function of the new gates that kept on changing. As an example, since software started eating the world (Andreessen, 2011), we are witnessing the spread of the software engineer, literally and figuratively through the web of the world.In a world increasingly networked, we needed to optimize them to channel trust, to translate, filter noise and scale decisions. But as we measured their performance by the value they take from their connections, we pressured them to develop methods and behaviors optimized for speed.As we pass through these early AI agentic days, the gatekeeper is perhaps in a tricky place. They can listen better, translate better, reduce errors through the gate and get more value from connections. They can prompt their way towards success, perhaps even consider taking themselves out of the loop. On the risky side, if they become overly controlling, they may consciously or unconsciously lie with numbers, which in turn can also take them out of the loop.There is a reason for a pause, a moment to revisit the role of the gatekeeper. It starts with this: what if we mechanized ourselves in ways that are not doing us good? As machines, and valued by performance, our cracks are showing. We have a lot to lose, especially now. But as humans, there’s something to take. Better than waiting to be disrupted is the opportunity to disrupt what we have made of ourselves.Where gatekeepers liveThey don’t have time for all kinds of "shit”, ideas. It doesn’t matter because if bad things pass through, the entire universe collapses. Keep in mind, though, that for gatekeepers under pressure, "the end of the universe" might in real life be as simple as a missing number in a status report.The space where they live is generally a bottleneck. He, the gatekeeper, deals with the necessity of filtering out bad input—bad decisions, bad collaborators, bad buyers, you name it—and selecting the good. As their world became networked, and the stream of inbound information scaled, they were prompted to adapt.In open source projects, such as Mozilla or Linux projects, gatekeepers dealt with an increasing stream of information from contributors willing to participate. Over time, the way the best contributors get in shapes the criteria that select future collaborators. From studies that dug into the caves of open source projects, we can get an idea of the dynamics between the underlying mechanisms of selection and the behaviors and methods of communication:Several studies provide more evidence towards technical expertise and its importance for successful contributions by measuring software artifacts developed by newcomers.For instance, some studies reported that sending messages or patches to a mailing list or issue tracker presenting previous technical skills can benefit the newcomer when joining.Stol et al. [PS16] evidenced that “when newcomers mentioned that they had already tried some options to fix their problem and have put efforts to look for a solution in the forums … then the responders were quick to respond and were very helpful.”They explain it as “a message of legitimacy from the newcomer along the lines of ‘I have done my homework, can I get some help now.’(Steinmacher et al., 2015)This is, of course, just a slice of how the system works. But it gives us a hint that the bigger the stream, the greater the pressure to manage the inbound information, faster and more efficiently.Implied in this situation is the fact that not everything can be accommodated. Many newcomers felt hurt, so they went away. Other newcomers found ways to work around it. They had to innovate. At the next stop, let’s look at situations when the gatekeeper is under pressure.Pressure through the gatesThe opening chapter of Silicon Valley began when William Shockley, co-inventor of the transistor, moved to Mountain View, California, near his mother. Shockley Semiconductor Laboratory was set to mass-produce and commercialize the transistor technology - the major invention of the 20th century and the foundational building block of modern electronics and computers.This was Mr. Shockley opening the gates to what, about 15 years later, would be called Silicon Valley. According to Robert Noyce, who was invited to work with Shockley and would later co-found Intel, "it was like talking to God" (PBS, 2013, 11:11). I can’t resist the irony of seeing Shockley himself as a controller, amplifying or switching like a transistor:A bipolar transistor allows a small current injected at one of its terminals to control a much larger current between the remaining two terminals, making the device capable of amplification or switching.(Wikipedia contributors, n.d., para. 1)Shockley was starting to face the 10x force, as the semiconductor landscape began to shift beneath him. The historical episode of “The Traitorous Eight,” gives a sense of the magnitude of change about to unfold. By September 1957, after just about one year at Shockley Semiconductor, eight of Shockley’s scientists and engineers decided to leave.The tip of the iceberg makes the big news when the ship sinks: On one side, eight rebels as seen by Shockley. On the other side, these eight portrayed Shockley’s style as unsuitable for a manager—a micromanager, impatient, and even paranoid about threats. But what unfolded next, after the traitorous eight founded Fairchild Semiconductor, shows the bottom half of the iceberg:By 1977, according to Fairchild, Fairchildren, and The Family Tree of Silicon Valley, about 126 semiconductor companies could be traced directly to Fairchild. Gordon Moore, one of the founders who later cofounded Intel, recounted how the movement mobilized engineers to embrace entrepreneurship:It seemed like every time we had a new product idea we had several spin-offs. Most of the companies around here even today can trace their lineage back to Fairchild. It was really the place that got the engineer entrepreneur really moving.(Computer History Museum, 2016, 1:48)This shift rebooted Silicon Valley from a culture where leaving was hard to one where leaving to found a spin-off felt exciting, if not required. This is, in part, the rise of the engineer as an entrepreneur. Bill Draper, who helped build the venture capital ecosystem supporting technology companies in the Bay Area, recounts how Fred Terman—the Dean of Engineering who later became Stanford’s provost—closed the gap between engineers and venture capitalists. Literally and figuratively, Stanford University provided the ground for what came next:I went to Yale, and at Yale we had an engineering department that was equal to Stanford’s. I graduated in 1950, and a guy named Witt Griswold taught me American history.Halfway through, he was asked to be president of Yale, which he did. And when he got in, he was a great, he’d been a great history teacher. But he made some huge mistakes as president of Yale. One of the big ones was that he said, ‘You know, engineering, that’s like dentistry. We don’t teach dentistry; we shouldn’t be teaching engineering. We teach biology, chemistry and physics and all the sciences. We shouldn’t teach engineering.’ So he cut back the engineering department quite rapidly.At the same time, Fred Terman beefed it up. He became the provost, had a strong position on the budget, and beefed up the engineering department at Stanford at exactly the same time.(Stanford, 2016, 8:23–9:40)The gatekeeper measured by outputConsider how the technological gatekeeper started as a translator, one who brought in and understood a new technology, and found a place as the go-to person when it came to applicability:There are, of course, possible measures which can be applied to reduce the There can exist in an R&D laboratory certain key individuals who are capable of effectively bridging the organizational boundary impedance and who provide the most effective entry point for ideas into the lab. These gatekeepers will be characterized in three ways:a. They will be the people to whom others in the lab most frequently turn for technical advice and consultation.b. They, themselves, will be more exposed (than others in the lab) to such formal media as the scientific and technological literature.c. In addition to exposure to formal media, the gatekeepers will maintain a greater degree of informal contact with members of the scientific-technological community outside of their own laboratory.(Allen, 1966)As gatekeepers evolved beyond the go-to role around knowledge, they embraced leadership roles. In the following chart, Jonathan Nightingale (Nightingale & Nightingale, 2017) suggests a path for a leader to improve as a function of conscious communication. Jonathan says that if you are an extreme huggybear, you need to move towards holding your team accountable. And on the other end, if you are radically results-oriented, you become a better leader by increasing sensitivity to relationship focus:This view aligns with Andy Grove’s idea of managerial leverage in High Output Management: the output of a manager is the output of the organization he or she manages. Andy’s viewpoint also aligns with the idea that leadership comes from enabling a team to communicate effectively and clearly:I have to confess that the information most useful to me, and I suspect most useful to all managers, comes from quick, often casual verbal exchanges. This usually reaches a manager much faster than anything written down, and usually the more timely the information, the more valuable it is.So why are written reports necessary at all? They obviously can’t provide timely information. What they do is constitute an archive of data, help to validate ad hoc inputs, and catch, in safety-net fashion, anything you have missed.But reports also have another totally different function: as they are formulated and written, the author is forced to be more precise than he might be verbally. Hence, their value stems from the discipline and the thinking the writer is forced to impose upon himself as he identifies and deals with verbal spots in his presentation.Reports are more a medium of self-discipline than a way to communicate information. Writing the report is important. Reading it often is not.(Grove, 2015, p. 48)The context window of the gatekeeperMy first idea about optimization in communication I learned from Professor D. Mr D was different, the only professor who joined us in the lab. When he was mad at the computer he would say the Brazilian word for shit three times, and beat the keyboard.One day he told us of his approach to dealing with the lazy students. He would take all the tests and give them 3 (out of 10). And wait. At some point a few students would show up in his office to complain. For them, he would start looking at the test to give a (proper) grading. I valued his honesty and understood his problem as a problem of dealing with time and energy.This was when I started forming the idea that people in certain positions adopting the filter-first think-later mode.What is the elevator pitch if not a little bit of the same thing as the founder approaches the garden of investors? Or the 50-minute presentation, if they pass through the elevator. A16z investor Marc Andreessen sheds light on how we miss the point. He starts by asking: why do we want the founder looking like an idiot presenting slides for 50 minutes? Why is it not a casual conversation? Why not have upfront research? “That is yet another test,” he says. All of that is to see how good a founder is at selling the idea, to customers, for example, who won’t give them any money, to engineers who have 20 other job offers to consider, and so on.We love when somebody walks in and has a compelling pitch so we can give them a check, that is a successful day for us.In contrast, every other pitch you ever gonna make is going to somebody who is going to be much worse than us.(Y Combinator, 2016, 11:34)That is how approaching the investor’s garden looks like: 3 min, 50 min, let’s see. Filter-first think-later had to rise as a result of gatekeepers in this domain optimizing for speed, managing time and resources in front of selected participants.The idea referred to as elimination by aspects was introduced in the work by Yin and Luo (2017) to help explain how accelerators select startups. They found that accelerators use “rejection first” as a way to trim alternatives before further evaluation. Essentially, they wrote, it is a process “to conserve cognitive effort” where a “set of criteria to reject startups as quickly as possible and trim the number of investment alternatives that they need to evaluate for funding”. According to the authors, the findings helped accelerator managers be conscious of their own subconscious preferences, rationales, and biases:To conserve cognitive efforts, investors implicitly used a parsimonious set of criteria to reject startups as quickly as possible and trim the number of investment alternatives that they need to evaluate for funding.However, investors often are not conscious of their preferences for certain evaluation criteria.The managers of a top accelerator are likely to experience similar cognitive capacity challenges when attracting many applications, and then they adopt a similar decision heuristic and process.(Yin & Luo, 2017, p. 26)The following is an example from the idea of Elimination by Aspects:In contemplating the purchase of a new car, for example, the first aspect selected may be automatic transmission: this will eliminate all cars that do not have this feature.Given the remaining alternatives, another aspect, say a $3000 price limit, is selected and all cars whose price exceeds this limit are excluded. The process continues until all cars but one are eliminated.(Tversky, 1972)To demonstrate how powerful the model appeals to humans, the paper from Tversky showed how the logic was used in a TV commercial. The commercial starts with two dozen eggs, and one walnut, representing companies offering training in computer programming in the San Francisco area:“How many of these schools have on-line computer facilities for training?”The announcer removes several eggs.“How many of these schools have placement services that would help find you a job?”The announcer removes some more eggs.“How many of these schools are approved for veterans’ benefits?”This continues until the walnut alone remains.The announcer cracks the nutshell, which reveals the name of the company and concludes:“This is all you need to know in a nutshell.”(Tversky, 1972)Living with it, the function withinDuring World War II, housewives as gatekeepers were considered in a qualitative study of decision-making in the household food channel. Among various factors influencing decisions the study reminds us of the idea of living with a decision once it is made:Let us assume that the housewife decides to buy an expensive piece of meat: the food passes the gate. Now the housewife will be very eager not to waste it.The forces formerly opposing each other will now both point in the same direction: the high price that tended to keep the expensive food out is now the reason why the housewife makes sure that through all the difficulties the meat gets safely to the table and is eaten.(Lewin, 1943, p. 37)The feedback loopWhat we originally wanted from them was interpretation, more than simple translation. But under pressure and obsessed with performance, we overwhelmed them. We pushed them too hard and made them, in many ways, like machines.Now the question goes back to a world where pills solve most of our problems. We have AI pills, agentic pills, you name it. In this world, where everything seems so blue, blue is the answer. So listen up: here is a story of a little guy who lives in a blue world (Wikipedia contributors, n.d.).With the blue pill, we can use agents to perform better, to keep getting things done the way we do, perhaps faster. But I hope that there is a red world, far away like Mars, where we can first look at ourselves from a distance, and then give feedback about the loop we are living in.ReferencesLewin, K. (1943). Forces behind food habits and methods of change. In National Research Council, The problem of changing food habits: Report of the Committee on Food Habits 1941–1943 (pp. 35–65). National Academy of Sciences.Andreessen, M. (2011, August 20). Why software is eating the world. Andreessen Horowitz. https://a16z.com/why-software-is-eating-the-world/Steinmacher, I., Graciotto Silva, M. A., Gerosa, M. A., & Redmiles, D. F. (2015). A systematic literature review on the barriers faced by newcomers to open source software projects. Information and Software Technology, 59, 67–85. https://doi.org/10.1016/j.infsof.2014.11.001Wikipedia contributors. (n.d.). Bipolar junction transistor. In Wikipedia. Retrieved September 11, 2026, from https://en.wikipedia.org/wiki/Bipolar_junction_transistorAllen, T. J. (1966). Managing the flow of scientific and technological information [Doctoral dissertation, Massachusetts Institute of Technology].PBS. (2013, February 1). AMERICAN EXPERIENCE | Silicon Valley Chapter 1 | PBS [Video]. YouTube. https://youtu.be/LcOoQP7nhl4Computer History Museum. (2016, June 22). Gordon Moore—Beginnings of entrepreneurship [Video]. YouTube. https://youtu.be/vH_cFnUx8kgStanford. (2016, May 31). Bill Draper: The rise of Sand Hill Road and Silicon Valley [Video]. YouTube.Nightingale, J., & Nightingale, M. (2017). How fcked up is your management?: An uncomfortable conversation about modern leadership*. Raw Signal Press.Grove, A. S. (2015). High output management. Vintage Books.Y Combinator. (2016, October 25). Marc Andreessen at Startup School SV 2016 [Video]. YouTube. https://www.youtube.com/watch?v=NEOR0AJsziEYin, B., & Luo, J. (2017). How do accelerators select startups? Shifting decision criteria across stages [Manuscript]. SSRN. https://doi.org/10.2139/ssrn.2735465Tversky, A. (1972). Elimination by aspects: A theory of choice. Psychological Review, 79(4), 281–299. https://doi.org/10.1037/h0032955Wikipedia contributors. (n.d.). Blue (Da Ba Dee). In Wikipedia. Retrieved September 16, 2026, from https://en.wikipedia.org/wiki/Blue_(Da_Ba_Dee)