Semi‑detached projects sit between organic and embedded. This classification matters because the same size system can require very different effort depending on constraints and novelty. It was derived from a study of 63 projects, which gives it unusually strong empirical roots for its time. Diversification does not eliminate the risk of experiencing investment losses. Investors should consider their investment objectives and risks carefully before investing. Keep in mind that while diversification may help spread risk, it does not assure a profit or protect against loss in a down market. Our analyst team just revealed what they believe are the 10 best stocks to buy right now, when you join Stock Advisor.COCOMO isn’t perfect, but it’s honest about uncertainty and structured enough to be improved.I do a short architecture review early and adjust cost drivers based on decisions like data consistency, latency targets, or cloud constraints.Let’s say the EAF works out to 0.85 (which is plausible if tools and capabilities offset the complexity).Some in the industry are frustrated, saying opponents are spreading falsehoods about data centers — such as polluting water and air — and are difficult to overcome.Instead of applying one set of cost drivers to the entire project, it assesses each development phase separately and sums the results.I also use this breakdown to assess risk. Nearly all of the world's largest oil traders expect a state of oil oversupply in 2026. Falling oil and gasoline prices have begun to push energy stocks lower. Brent, the benchmark for oil from Europe, Africa, and the Middle East, is now priced at about $60 a barrel, also $15 below where it was at the beginning of 2025. West Texas Intermediate, the type of oil extracted from oilfields in the U.S., is currently trading at about $57 a barrel, $15 below where it started the year. As you might imagine, that glut has pushed oil prices down in a big way. Mike Petak of Spring City gestures while speaking to East Vincent Township supervisors in opposition to a data center proposal at the former Pennhurst state hospital grounds, Dec. 17, 2025, in Spring City, Pa. (AP Photo/Marc Levy) All information and data on the website is for reference only and no historical data shall be considered as the basis for judging future trends. No content on the Webull Financial LLC website shall be considered as a recommendation or solicitation for the purchase or sale of securities, options, or other investment products. When to Use COCOMO—and When Not To That can shift effort into later phases, especially integration and test. The real benefits tend to show up in implementation and unit testing, not in requirements or integration. Potential 10–15% effort reductionThat makes the estimate actionable rather than purely descriptive. Use saved searches to filter your results more quickly I adjust cost drivers for experience and complexity, and I explicitly add learning buffers. If the team has never built anything similar, effort grows. Early in a project, size is a range, not a point. When I have enough information, I use a detailed model to split effort across phases. 4) Convert the effort into a phase‑based plan to avoid under‑resourcing design or testing. Here’s a simple, runnable example that calculates basic and intermediate effort. AI can produce code fast, but without strong review and testing practices, defects can increase. I use it when I need a fast estimate on a sketchy scope or as a quick sanity check against more detailed approaches. The coefficients a and b vary by project type (organic, semi‑detached, embedded). If you’re writing software that must meet stringent certification or real‑time guarantees, you should treat it as embedded even if the codebase doesn’t look huge. COCOMO assumes the highest effort multiplier here. The problem is well understood, and the team has relevant experience. related products There are currently 1.4 billion barrels of oil on the water -- i.e., oil being shipped to a port or stored and waiting for a buyer. Our analyst team just revealed what they believe are the 10 best stocks to buy right now, when you join Stock Advisor. As a developer, Git‘s version control system helps you track changes to your code over time. If the numbers are far apart, that’s not a failure—it’s the start of a better conversation. Flight control or medical deviceThat size range is a historical artifact; modern systems can have fewer lines of code but more integration and operational complexity.3) Underestimating integration and data workInvestors should consider their investment objectives and risks carefully before investing.West Texas Intermediate, the type of oil extracted from oilfields in the U.S., is currently trading at about $57 a barrel, $15 below where it started the year.I produce a “likely” estimate and then a low and high scenario.Aggressive schedules increase cost, often sharply. The Intermediate model includes multiple cost drivers. If your code generator doubles output, it doesn’t mean effort halves. Instead of applying one set of cost drivers to the entire project, it assesses each development phase separately and sums the results. Intermediate COCOMO People listen during an East Vincent Township supervisors meeting where an agenda item involved a data center proposal at the former Pennhurst state hospital grounds, Dec. 17, 2025, in Spring City, Pa. (AP Photo/Marc Levy) People opposed to a data center proposal at the former Pennhurst state hospital grounds attend an East Vincent Township supervisors meeting, Dec. 17, 2025, in Spring City, Pa. (AP Photo/Marc Levy) People sign in and head into an East Vincent Township supervisors meeting where an agenda item involved a data center proposal at the former Pennhurst state hospital grounds, Dec. 17, 2025, in Spring City, Pa. (AP Photo/Marc Levy) People opposed to a data center proposal at the former Pennhurst state hospital grounds talk during a break in an East Vincent Township supervisors meeting, Dec. 17, 2025, in Spring City, Pa. (AP Photo/Marc Levy) The Structure of the COCOMO Model I always communicate estimates as ranges and conditions.With a team of 15 people, 36 person‑months is about 2.4 months.Semi‑detached projects sit between organic and embedded.People opposed to a data center proposal at the former Pennhurst state hospital grounds attend an East Vincent Township supervisors meeting, Dec. 17, 2025, in Spring City, Pa. (AP Photo/Marc Levy)The model gives you a structured way to discuss risk, scope, and team capability.This gives stakeholders a risk‑aware view rather than a single number that will be interpreted as a promise.The value of securities may fluctuate and as a result, clients may lose more than their original investment.Communities across the United States are reading about — and learning from — each other’s battles against data center proposals that are fast multiplying in number and size to meet steep demand as developers branch out in search of faster connections to power sources.Margin trading increases risk of loss and includes the possibility of a forced sale if account equity drops below required levels.In Indiana alone, Gustafson counted more than a dozen projects that lost rezoning petitions. A $0.50 per contract fee applies for certain index options and a $0.10 per contract fee applies for oversized option orders. Free trading of stocks, ETFs, and options refers to $0 commissions for Webull Financial LLC self-directed individual cash or margin brokerage accounts and IRAs that trade U.S. listed securities via mobile devices, desktop or website products. Leverage carries a high level of risk and is not suitable for all investors. Margin trading increases risk of loss and includes the possibility of a forced sale if account equity drops below required levels. Either way, the model provides a disciplined way to explore those questions. If the estimate feels too high, don’t just throw it out—use it to examine your assumptions. 5) Re‑estimate when scope or architecture changes. I do a short architecture review early and adjust cost drivers based on decisions like data consistency, latency targets, or cloud constraints. If it’s wildly off, I revisit assumptions rather than discard the model. In those cases, I rely more on time‑boxed discovery and incremental planning rather than a formal cost model. I keep a lightweight history of delivered projects to adjust coefficients and EAFs. 3) Underestimating integration and data work That lets me provide a realistic effort range and avoid false precision. I always estimate a low, likely, and high KLOC. This is especially useful in large programs where multiple teams share components. This phase‑based breakdown is essential when you need to align resources or coordinate multiple teams. Futures and other assets held outside the securities account are not covered. (ii) For securities accounts that are fully-disclosed to the clearing firm, Apex has purchased an additional insurance policy. Please read the Risk Disclosure Statement and other relevant Futures Disclosures located at /fcm-disclosures prior to trading futures products. Futures and futures options trading involves substantial risk and is not suitable for all investors. Suppose you’re building a new customer insights platform with multiple services, a data pipeline, and a dashboard.The coefficients a and b vary by project type (organic, semi‑detached, embedded).These factors remind stakeholders that estimates aren’t just about size.AI can produce code fast, but without strong review and testing practices, defects can increase.Data Center Watch, a project of 10a Labs, an AI security consultancy, said it is seeing a sharp escalation in community, political and regulatory disruptions to data center development.It’s not a perfect mapping, but it preserves the idea that effort isn’t uniform across the lifecycle.Options trading entails significant risk and is not appropriate for all investors. I adjust tool and capability drivers slightly, but I never halve estimates just because we use AI. These choices often dominate the cost. This gives stakeholders a risk‑aware view rather than a single number that will be interpreted as a promise. I produce a “likely” estimate and then a low and high scenario. I account for AI assistance by adjusting the TOOL driver and occasionally the PCAP driver, but I never assume it halves the effort. If the model consistently overshoots or undershoots, I adjust coefficients. Effort rises and risk of defects increases When I present a COCOMO estimate, I keep it grounded and honest. You don’t need a massive dataset; even five to ten projects can help calibrate a and b coefficients to your organization. If data is central, I classify the project as semi‑detached or embedded, even if the app code is straightforward. If the model says system design is 15% of effort, I can estimate how many weeks that represents and plan staffing accordingly. If the project has a heavy integration surface, I push more effort into integration and test. Intermediate COCOMO extends the basic model with a set of cost drivers, often called Effort Adjustment Factors (EAF). Each version adds more fidelity by accounting for additional cost drivers and by splitting the project into phases. In these projects, a single requirement change can trigger deep redesigns, and the environment imposes hard constraints on architecture choices. Embedded projects are complex, tightly constrained, and often tied to strict performance, safety, or hardware limits. I see most enterprise projects land in this category. The scope is roughly 120 KLOC when you include backend services, data jobs, and UI. Suppose you’re building a new customer insights platform with multiple services, a data pipeline, and a dashboard. They’re about the environment and the team. Aggressive schedules increase cost, often sharply. If your team hasn’t adopted these tools, assume average. The past performance of a security, or financial product does not guarantee future results or returns. The value of securities may fluctuate and as a result, clients may lose more than their original investment. All investments involve risk, and not all risks are suitable for every investor. Options trading entails significant risk and is not appropriate for all investors. SIPC and Excess SIPC Protections do not protect against a loss in the market value of securities.SIPC is a non-profit, membership corporation funded by broker-dealers that are members of SIPC. It seems that the biggest producers of oil are now bracing for it. Energy Information Administration's latest outlook says rising inventories in 2026 will put downward pressure on oil prices in coming months, with Brent oil falling to $55 in the first quarter and remaining there through the end of the year. The International Energy Agency forecasts that global oil supplies will outpace demand by more than 3.8 million barrels a day next year, which would be a record mismatch between supply and demand for petroleum. With a team of 15 people, 36 person‑months is about 2.4 months. Suppose your updated estimate is 300 person‑months and you want to plan staffing and milestones. In both cases, I keep the conversion in a small table so that the team can see where it comes from. This keeps the estimate aligned with reality and prevents surprise schedule slips. Product Detail And lower oil prices are good for economic growth everywhere -- except in countries that are highly dependent on oil exports. Growing hostilities between the West and Russia could send prices higher again on concerns about supply, while a ceasefire or end to the Russia-Ukraine war would alleviate such concerns and send prices even lower by ending sanctions on Russian oil. To be sure, the price of oil and the profitability of oil companies are not wholly dependent on supply and demand. A growing number of proposals are going down in defeat, sounding alarms across the data center constellation of Big Tech firms, real estate developers, electric utilities, labor unions and more. But as more people hear about a data center coming to their community, once-sleepy municipal board meetings in farming towns and growing suburbs now feature crowded rooms of angry residents pressuring local officials to reject the requests. In many cases, municipal boards are trying to figure out whether energy- and water-hungry data centers fit into their zoning framework. I maintain a table of completed projects with size, actual effort, and key drivers. In my experience, effort correlates more with cognitive complexity and integration risk than raw KLOC. Flight control or medical deviceThat size range is a historical artifact; modern systems can have fewer lines of code but more integration and operational complexity. Organic projects are small to medium in size with relatively low complexity. No content on the Webull Financial LLC website shall be considered as a recommendation or solicitation for the purchase or sale of securities, options, or other investment products.In both cases, I keep the conversion in a small table so that the team can see where it comes from.That amounts to two-thirds of the projects it was tracking.The Intermediate model includes multiple cost drivers.I use it when I need a fast estimate on a sketchy scope or as a quick sanity check against more detailed approaches.As a developer, Git‘s version control system helps you track changes to your code over time.People opposed to a data center proposal at the former Pennhurst state hospital grounds talk during a break in an East Vincent Township supervisors meeting, Dec. 17, 2025, in Spring City, Pa. (AP Photo/Marc Levy) COCOMO classifies projects into three broad categories based on size, complexity, and environment. Yet estimation is still hard because the risk lives in integration, requirements ambiguity, and organizational friction—factors COCOMO tries to model, at least indirectly. Still, data center allies say they are urging developers to engage with the public earlier in the process, emphasize economic benefits, sow good will by supporting community initiatives and talk up efforts to conserve water and power and protect ratepayers. That is, lower oil prices force some producers to shut down projects, reduce investments in new sources, or even go out of business. If you’re leading a team in 2026, I’d encourage you to keep a small estimation history, calibrate your coefficients, and revisit estimates as the project evolves. But as more people hear about a data center coming to their community, once-sleepy municipal board meetings in farming towns and growing suburbs now feature crowded rooms of angry residents pressuring local officials to reject the requests.Free trading of stocks, ETFs, and options refers to $0 commissions for Webull Financial LLC self-directed individual cash or margin brokerage accounts and IRAs that trade U.S. listed securities via mobile devices, desktop or website products.Each version adds more fidelity by accounting for additional cost drivers and by splitting the project into phases.I’ve seen high‑capability teams deliver 30–40% faster than average on the same scope, but only when the problem is understood and the team is aligned.In those cases, I rely more on time‑boxed discovery and incremental planning rather than a formal cost model.If your team leans heavily on AI, I recommend increasing the integration/test phase allocation unless you have strong automated test coverage and architecture guardrails.The model predicts effort, cost, and schedule from a project’s size and a set of drivers that reflect complexity, constraints, and team capability. Large datasets affect infrastructure costs and development time because data modeling and testing scale poorly.It seems that the biggest producers of oil are now bracing for it.If your code generator doubles output, it doesn’t mean effort halves.The key is to keep the model honest.COCOMO (Constructive Cost Model) is a procedural software cost estimation model proposed by Barry Boehm in 1981.If the project has a heavy integration surface, I push more effort into integration and test.Mike Petak of Spring City gestures while speaking to East Vincent Township supervisors in opposition to a data center proposal at the former Pennhurst state hospital grounds, Dec. 17, 2025, in Spring City, Pa. (AP Photo/Marc Levy) The team has mixed experience with the domain, and there are moderate constraints around data privacy and latency. These factors remind stakeholders that estimates aren’t just about size. Modern toolchains, CI/CD, static analysis, and AI assistance can reduce effort—if the team knows how to use them. Even in cloud systems, latency and throughput requirements can dominate engineering effort. Large datasets affect infrastructure costs and development time because data modeling and testing scale poorly. That’s a practical choice from the 1980s, but it’s not the only way to estimate size today. I use size as a proxy, but I also adjust the project class based on architecture, external dependencies, and compliance requirements. If your team is seasoned, the estimate is mainly driven by size. If you change size or project class, you can see how your estimate changes. COCOMO (Constructive Cost Model) is a procedural software cost estimation model proposed by Barry Boehm in 1981. That’s where a cost estimation model like COCOMO earns its keep. In Hermantown, a suburb of Duluth, Minnesota, a proposed data center campus several times larger than the Mall of America is on hold amid challenges over whether the city’s environmental review was adequate. Some in the industry are frustrated, saying opponents are spreading falsehoods about data centers — such as polluting water and air — and are difficult to overcome. In an October securities filing, it listed its operational risks as including “community opposition, local moratoriums, and hyper-local dissent that may impede or delay infrastructure development.” And lower oil prices are good for economic growth everywhere -- except in countries that are highly dependent on oil exports.What makes COCOMO valuable isn’t that it’s always correct; it’s that it exposes your assumptions.In an October securities filing, it listed its operational risks as including “community opposition, local moratoriums, and hyper-local dissent that may impede or delay infrastructure development.”Yet estimation is still hard because the risk lives in integration, requirements ambiguity, and organizational friction—factors COCOMO tries to model, at least indirectly.” into “how much time, effort, and money will it take? If it’s a greenfield build with a strong architecture, I shift more to design. This helps me plan staffing and identify phase risks. That is the real value of the model. A team of 12 could do that in 24 months, or a team of 20 in around 15 months. If leadership wants a hard date, I show the assumptions and explain the risks of compressing the schedule. I always communicate estimates as ranges and conditions. The model is a planning tool, not a contract. COCOMO coefficients are derived from historical data. Data pipelines, API integration, and migration are notorious time sinks. The key is to keep the model honest. COCOMO traditionally uses KLOC as the size metric. It’s not a perfect mapping, but it preserves the idea that effort isn’t uniform across the lifecycle. For some people angry over steep increases in electric bills, their patience is thin for data centers that could bring still-higher increases. In Indiana alone, Gustafson counted more than a dozen projects that lost rezoning petitions. That amounts to two-thirds of the projects it was tracking. Andy Cvengros, who helps lead the data center practice at commercial real estate giant JLL, counted seven or eight deals he’d worked on in recent months that saw opponents going door-to-door, handing out shirts or putting signs in people’s yards. The model predicts effort, cost, and schedule from a project’s size and a set of drivers that reflect complexity, constraints, and team capability. AI code assistants can reduce effort on well‑specified tasks, but they don’t eliminate integration risk or requirements churn. The model assumes effort grows faster than size because coordination costs and technical risk are higher. Maybe your size estimate is too big, or maybe you’re underestimating integration complexity. This makes the size estimate more consistent across languages. For example, if your last three projects averaged 0.8 KLOC per story point, and the new project is 100 points, you can start with 80 KLOC. If you want to go beyond KLOC, you can adapt the model by mapping other size metrics. When requirements stabilize or architecture changes, I rerun the model. At the same time, lower oil prices increase the demand for oil and oil products like gasoline, because people use them more when they're cheaper. And of course, there's the somewhat accurate observation that the cure for low oil prices is low prices. They're also not good for oil companies and their shareholders. COCOMO isn’t perfect, but it’s honest about uncertainty and structured enough to be improved. If integration and test is 20% but you’re integrating with five external systems, I increase that allocation. I also use this breakdown to assess risk. What makes COCOMO valuable isn’t that it’s always correct; it’s that it exposes your assumptions. ” into “how much time, effort, and money will it take? Documents revealing the extent of the project emerged days before a city rezoning vote in October. The project would have funded half the city’s budget and developers promised environmentally friendly features. I’ve seen high‑capability teams deliver 30–40% faster than average on the same scope, but only when the problem is understood and the team is aligned. Real‑time performance or memory limitations introduce non‑linear cost. This driver is often underappreciated in estimates. Complex algorithms, distributed coordination, or heavy domain logic increases effort. If the system is safety‑critical or must achieve very high uptime, cost grows fast. Developers pulled a project off an October agenda in the Charlotte suburb of Matthews, North Carolina, after Mayor John Higdon said he informed them it faced unanimous defeat. “The thing is you could have power to a site and it’s futile because you might not get the zoning. “Because that’s where it’s literally going, is in my backyard.” There is always the potential of losing money when you invest in securities or other financial products. You’ll get more accurate forecasts over time, and you’ll build trust with stakeholders because your estimates will be grounded in data and transparent reasoning. If you need a credible estimate, I recommend starting with COCOMO as a baseline and then calibrating with your own data. 2) Choose the project class based on constraints and novelty, not just size. If your team leans heavily on AI, I recommend increasing the integration/test phase allocation unless you have strong automated test coverage and architecture guardrails. That’s still too large for a team of 10 over a year. Let’s say the EAF works out to 0.85 (which is plausible if tools and capabilities offset the complexity). It’s telling us the project is non‑trivial, but it doesn’t reflect our actual tooling or modern productivity. I’d classify this as a semi‑detached project. The model gives you a structured way to discuss risk, scope, and team capability. It can generate boilerplate, translate requirements into code, and automate tests. So planning and requirements would take roughly 10 weeks with that team size. If your team uses story points and has historical data, estimate average KLOC per point. At its core, COCOMO assumes a relationship between size (often expressed in KLOC, thousand lines of code) and effort (person‑months). I’ll explain the project categories, the model structure, and the math behind the estimates, then show how I use it alongside real metrics and AI‑assisted planning. Data Center Watch, a project of 10a Labs, an AI security consultancy, said it is seeing a sharp escalation in community, political and regulatory disruptions to data center development. Communities across the United States are reading about — and learning from — each other’s battles against data center proposals that are fast multiplying in number and size to meet steep demand as developers branch out in search of faster connections to power sources.