Is Open Source Doomed?
September 6, 2024
Open source cheerleaders may need to find a new team to route for. Web developer and blogger Baldur Bjarnason describes “The Slow Evaporation of the Free/Open Source Surplus.” He notes he is joining a conversation begun by Tara Tarakiyee with the post, Is the Open Source Bubble about to Burst? and continued by Ben Werdmuller.
Bjarnason begins by specifying what has made open source software possible up until now: surpluses in both industry (high profit margins) and labor (well-paid coders with plenty of free time.) Now, however, both surpluses are drying up. The post lists several reasons for this. First, interest rates remain high. Next, investment dollars are going to AI, which “doesn’t really do real open source.” There were also the waves of tech layoffs and cost-cutting after post-pandemic overspending. Severe burnout from a thankless task does not help. We are reminded:
“Very few FOSS projects are lucky enough to have grown a sustainable and supportive community. Most of the time, it seems to be a never-ending parade of angry demands with very little reward.”
Good point. A few other factors, Bjarnason states, make organizations less likely to invest in open source:
- Why compete with AWS or similar services that will offer your own OSS projects at a dramatically lower price?
- Why subsidise projects of little to no strategic value that contribute anything meaningfully to the bottom-line?
- Why spend time on OSS when other work is likely to have higher ROI?
- Why give your work away to an industry that treats you as disposable?”
Finally, Bjarnason suspects even users are abandoning open source. One factor: developers who increasingly reach for AI generated code instead of searching for related open source projects. Ironically, those LLMs were trained on open source software in the first place. The post concludes:
Best case scenario, seems to me, is that Free and Open Source Software enters a period of decline. After all, that’s generally what happens to complex systems with less investment. Worst case scenario is a vicious cycle leading to a collapse:
- Declining surplus and burnout leads to maintainers increasingly stepping back from their projects.
- Many of these projects either bitrot serious bugs or get taken over by malicious actors who are highly motivated because they can’t relay on pervasive memory bugs anymore for exploits.
- OSS increasingly gets a reputation (deserved or not) for being unsafe and unreliable.
- That decline in users leads to even more maintainers stepping back.”
Bjarnason notes it is possible some parts of the Open Source ecosystem will not crash and burn. Overall, though, the outlook seems bleak.
Cynthia Murrell, September 6, 2024
Hey, Alexa, Why Does Amazon AI Flail?
September 5, 2024
This essay is the work of a dumb dinobaby. No smart software required.
Amazon has its work cut out for itself. The company has those pesky third-party vendors shipping “interesting” products to customers and then ignoring complaints. Amazon is on the radar of some legal eagles in the EU and the US. Now the company has found itself in an unusual situation: Its super duper smart software does not work. The fix, if the information in “Gen AI Alexa to Use Anthropic Tech After it Struggled for Words” with Amazon’s” is correct, is to use Anthropic AI technology. Hey, why not? Amazon allegedly invested $5 billion in the company. Maybe that implementation of Google technology will do the trick?
The mother is happy with Alexa’s answers. The weird sounds emitted from the confused device surprise her daughter. Thanks, MSFT Copilot. Good enough.
The write up reports:
Amazon demoed a generative AI version of Alexa in September 2023 and touted it as being more advanced, conversational, and capable, including the ability to do multiple smart home tasks with simpler commands. Gen AI Alexa is expected to come with a subscription fee, as Alexa has reportedly lost Amazon tens of billions of dollars throughout the years. Earlier reports said the updated voice assistant would arrive in June, but Amazon still hasn’t confirmed an official release date.
A year later, Amazon is punting and giving the cash furnace Alexa more brains courtesy of Anthropic. Will the AI wizards working on Amazon’s own AI have a chance to work in one of the Amazon warehouses?
Ars Technica says without a trace of irony:
The previously announced generative AI version of Amazon’s Alexa voice assistant “will be powered primarily by Anthropic’s Claude artificial intelligence models," Reuters reported today. This comes after challenges with using proprietary models, according to the publication, which cited five anonymous people “with direct knowledge of the Alexa strategy.”
Amazon has a desire to convert the money-losing Alexa into a gold mine, or at least a modest one.
This report, if accurate, suggests some interesting sparkles on the Bezos bulldozer’s metal flake paint; to wit:
- The two pizza team approach to technology did not work either for Alexa (the money loser) or the home grown AI money spinner. What other Amazon technologies are falling short of the mark?
- How long will it take to get a money-generating Alexa working and into the hands of customers eager for a better Alexa experience and a monthly or annual subscription for the new Alexa? A year has been lost already, and Alexa users continue to ask for the weather and a timer for cooking broccoli.
- What happens if the product, its integration with smart TV, and the Ring doorbell is like a Pet Rock? The fad has come and gone, replaced by smart watches and mobile phones? The answer: Collectibles!
Why am I questioning Amazon’s technology competency? The recent tie up between Microsoft and Palantir Technologies makes clear that Amazon’s cloud services don’t have the horsepower to pull government sales. When these pieces are shifted around, the resulting puzzle says, “Amazon is flailing to me.” Consider this: AI was beyond the reach of a big money outfit like Amazon. There’s a message in that factoid.
Stephen E Arnold, September 5, 2024
Uber Leadership May Have to Spend Money to Protect Drivers. Wow.
September 5, 2024
This essay is the work of a dumb dinobaby. No smart software required.
Senior managers — now called “leadership” — care about their employees. I added a wonderful example about corporate employee well being and co-worker sensitivity when I read “Wells Fargo Employee Found Dead in Her Cubicle 4 Days After She Clocked in for Work.” One of my team asked me, “Will leadership at that firm check her hours of work so she is not overpaid for the day she died?” I replied, “You will make a wonderful corporate leader one day.” Another analyst asked, “Didn’t the cleaning crew notice?” I replied, “Not when they come once every two weeks.”
Thanks, MSFT Copilot. Good enough given your filters.
A similar approach to employee care popped up this morning. My newsreader displayed this headline: “Ninth Circuit Rules Uber Had Duty to Protect Washington Driver Murdered by Passengers.” The write up reported:
The estate of Uber driver Cherno Ceesay sued the rideshare company for negligence and wrongful death in 2021, arguing that Uber knew drivers were at risk of violent assault from passengers but neglected to install any basic safety measures, such as barriers between the front and back seats of Uber vehicles or dash cameras. They also claimed Uber failed to employ basic identity-verification technology to screen out the two customers who murdered Ceesay — Olivia Breanna-Lennon Bebic and Devin Kekoa Wade — even though they opened the Uber account using a fake name and unverified form of payment just minutes before calling for the ride.
Hold it right there. The reason behind the alleged “failure” may be the cost of barriers, dash cams, and identity verification technology. Uber is a Big Dog high technology company. Its software manages rides, maps, payments, and the outstanding Uber app. If you want to know where your driver is, text the professional. Want to know the percentage of requests matched to drivers from a specific geographic point, forget that, gentle reader. Request a ride and wait for a confirmation. Oh, what if a pick up is cancelled after a confirmation? Fire up Lyft, right?
The cost of providing “basic” safety for riders is what helps make old fashioned taxi rides slightly more “safe.” At one time, Uber was cheaper than a weirdly painted taxi with a snappy phone number like 666 6666 or 777 7777 painted on the side. Now that taxis have been stressed by Uber, the Uber rides have become more expensive. Thanks to surge pricing, Uber in some areas is more expensive than taxis and some black car services if one can find one.
Uber wants cash and profits. “Basic” safety may add the friction of additional costs for staff, software licenses, and tangibles like plastic barriers and dash cams. The write up explains by quoting the legalese of the court decision; to wit:
“Uber alone controlled the verification methods of drivers and riders, what information to make available to each respective party, and consistently represented to drivers that it took their safety into consideration Ceesay relied entirely on Uber to match him with riders, and he was not given any meaningful information about the rider other than their location,” the majority wrote.
Now what? I am no legal eagle. I think Uber “leadership” will have meetings. Appropriate consultants will be retained to provide action plan options. Then staff (possibly AI assisted) will figure out how to reduce the probability of a murder in or near an Uber contractor’s vehicle.
My hunch is that the process will take time. In the meantime, I wonder if the Uber app autofills the “tip” section and then intelligently closes out that specific ride? I am confident that universities offering business classes will incorporate one or both of these examples in a class about corporate “leadership” principles. Tip: The money matters. Period.
Stephen E Arnold, September 5, 2024
What are the Real Motives Behind the Zuckerberg Letter?
September 5, 2024
Senior correspondent at Vox Adam Clarke Estes considers the motives behind Mark Zuckerberg’s recent letter to Rep. Jim Jordan. He believes “Mark Zuckerberg’s Letter About Facebook Censorship Is Not What it Seems.” For those who are unfamiliar: The letter presents no new information, but reminds us the Biden administration pressured Facebook to stop the spread of Covid-19 misinformation during the pandemic. Zuckerberg also recalls his company’s effort to hold back stories about Hunter Biden’s laptop after the FBI warned they might be part of a Russian misinformation campaign. Now, he insists, he regrets these actions and vows never to suppress “freedom of speech” due to political pressure again.
Naturally, Republicans embrace the letter as further evidence of wrongdoing by the Biden-Harris administration. Many believe it is evidence Zuckerberg is kissing up to the right, even though he specifies in the missive that his goal is to be apolitical. Estes believes there is something else going on. He writes:
“One theory comes from Peter Kafka at Business Insider: ‘Zuckerberg very carefully gave Jordan just enough to claim a political victory — but without getting Meta in any further trouble while it defends itself against a federal antitrust suit. To be clear, Congress is not behind the antitrust lawsuit. The case, which dates back to 2021, comes from the FTC and 40 states, which say that Facebook illegally crushed competition when it acquired Instagram and WhatsApp, but it must be top of mind for Zuckerberg. In a landmark antitrust case less than a month ago, a federal judge ruled against Google, and called it a monopoly. So antitrust is almost certainly on Zuckerberg’s mind. It’s also possible Zuckerberg was just sick of litigating events that happened years ago and wanted to close the loop on something that has caused his company massive levels of grief. Plus, allegations of censorship have been a distraction from his latest big mission: to build artificial general intelligence.”
So is it coincidence this letter came out during the final weeks of a severely close, high-stakes presidential election? Perhaps. An antitrust ruling like the one against Google could be inconvenient for Meta. Curious readers can navigate to the article for more background and more of Estes reasoning.
Cynthia Murrell, September 5, 2024
Accountants: The Leaders Like Philco
September 4, 2024
This essay is the work of a dumb dinobaby. No smart software required.
AI or smart software has roiled the normal routine of office gossip. We have shifted from “What is it?” to “Who will be affected next?” The integration of AI into work processes, however, is not a new thing. Most people don’t know or don’t recall that when a consultant could do a query from a clunky device like the Texas Instrument Silent 700, AI was already affecting jobs. Whose? Just ask a special librarian who worked when an intermediary was not needed to retrieve information from an online database.
A nervous smart robot running state-of-the-art tax software is sufficiently intelligent to be concerned about the meeting with an IRS audit team. Thanks, MSFT Copilot. How’s that security push coming along? Oh, too bad.
I read “Why America’s Most Boring Job Is on the Brink of Extinction.” I think the story was crafted by a person who received either a D or an F in Accounting 100. The lingo links accountants with being really dull people and the nuking of an entire species. No meteor is needed; just smart software, the silent killer. By the way, my two accountants are quite sporty. I rarely fall asleep when they explain life from their point of view. I listen, and I urge you to be attentive as well. Smart software can do some excellent things, but not everything related to tax, financial planning, and keeping inside the white lines of the quite fluid governmental rules and regulations.
Nevertheless, the write up cited above states:
Experts say the industry is nearing extinction because the 150-hour college credit rule, the intense entry exam and long work hours for minimal pay are unappealing to the younger generation.
The “real” news article includes some snappy quotes too. Here’s one I circled: “’The pay is crappy, the hours are long, and the work is drudgery, and the drudgery is especially so in their early years.’”
I am not an accountant, so I cannot comment on the accuracy of this statement. My father was an accountant, and he was into detail work and was able to raise a family. None of us ended up in jail or in the hospital after a gang fight. (I was and still am a sissy. Imagine that: An 80 year old dinobaby sissy with the DNA of an accountant. I am definitely exciting.)
With fewer people entering the field of accounting, the write up makes a remarkable statement:
… Accountants are becoming overworked and it is leading to mistakes in their work. More than 700 companies cited insufficient staff in accounting and other departments as a reason for potential errors in their quarterly earnings statements…
Does that mean smart software will become the accountants of the future? Some accountants may hope that smart software cannot do accounting. Others will see smart software as an opportunity to improve specific aspects of accounting processes. The problem, however, is not the accountants. The problem will AI is the companies or entrepreneurs who over promise and under deliver.
Will smart software replace the insight and timeline knowledge of an experienced numbers wrangler like my father or the two accountants upon whom I rely?
Unlikely. It is the smart software vendors and their marketers who are most vulnerable to the assertions about Philco, the leader.
Stephen E Arnold, September 4, 2024
Salesforces Disses Microsoft Smart Software
September 4, 2024
This essay is the work of a dumb dinobaby. No smart software required.
Senior managers can be frisky at times. A good example appears in the Fortune online service write up “Salesforce CEO Marc Benioff Says Microsoft Copilot Has Disappointed Many Customers.” I noted this statement in the article:
Marc Benioff said Microsoft’s Copilot AI hasn’t lived up to the hype…. unimpressive.
The old fish comparison works for smart software it seems. Thanks, MSFT Copilot. Good enough just not tastier.
Consider the number of organizations which use Microsoft and its smart software. Will those organizations benefit from “unimpressive” programs and services. What about the US government which might struggle to operate without Microsoft software. What if the US government operates in a way which delivers unimpressive performance? What about companies relying on Microsoft technology? Will these organizations struggle to deliver high-octane performance?
The article reported that the Big Dog of Salesforce opined:
“So many customers are so disappointed in what they bought from Microsoft Copilot because they’re not getting the accuracy and the response that they want,” Benioff said. “Microsoft has disappointed so many customers with AI.”
“Disappointed” — That’s harsh.
True to its rich history of business journalism, the article included a response from Microsoft, a dominant force in enterprise and consumer software (smart or otherwise). I noted this Microsoft comment:
Jared Spataro, Microsoft’s corporate vice president for AI at work, said in a statement to Fortune that the company was “hearing something quite different,” from its customers. The company’s Copilot customers also shot up 60% last quarter and daily users have more than doubled, Spataro added.
From Microsoft’s point of view, this is evidence that Microsoft is delivering high-value smart software. From Salesforce’s point of view, Microsoft is creating customers for Salesforce’s smart software. The problem is that Salesforce is not exactly the same type of software outfit as Salesforce. Nevertheless, the write up included this suggestive comment from the Big Dog of Salesforce:
“With our new Agentforce platform, we’re going to make a quantum leap for AI,” he said.
I like the use of the word “quantum.” It suggests uncertainty to me. I remain a bit careful when it comes to discussions of “to be” software. Marketing-type comments are far easier to create than a functional, reliable, and understandable system infused with smart software.
But PR and marketing are one thing. Software which does not hallucinate or output information that cannot be verified given an organization’s resources is different. Who cares? That’s a good question. Stakeholders, those harmed by AI outputs, and unemployed workers replaced by more “efficient” systems maybe?
Content marketing, sales hyperbole, and PR — The common currency of artificial intelligence makes life interesting.
Stephen E Arnold, September 4, 2024
Indifference or Carelessness: The Security Wrecks from Georgia Tech
September 4, 2024
DOJ Sues Georgia Tech for DOD-Related Cybersecurity Violations
The Justice Department takes cybersecurity standards for our military very seriously. Just ask Georgia Tech University. Nextgov/FCW reports, “DOJ Suit Claims Georgia Tech ‘Knowingly Failed’ to Meet Cyber Standards for DOD Contracts.” The suit began in 2022 with a whistleblower lawsuit filed by two members of the university’s cybersecurity compliance team. They did so under the DOJ’s Civil Cyber-Fraud Initiative. Now the DOJ has joined the fray. Reporter Edward Graham tells us:
“In a press release, DOJ alleged that the institutions committed numerous violations of the Department of Defense’s cybersecurity policy in the years prior to the whistleblower complaint. Among the most serious allegations was the claim that ‘Georgia Tech and [Georgia Tech Research Corporation] submitted a false cybersecurity assessment score to DOD for the Georgia Tech campus’ in December 2020. … The lawsuit also asserted that the Astrolavos Lab at Georgia Tech previously ‘failed to develop and implement a system security plan, which is required by DOD cybersecurity regulations.’ Once the security document was finally implemented in February 2020, the complaint said the university ‘failed to properly scope that plan to include all covered laptops, desktops and servers.’ Additionally, DOJ alleged that the Astrolavos Lab did not use any antivirus or antimalware programs on its devices until December 2021. The university reportedly allowed the lab to refuse the installation of the software ‘in violation of both federal cybersecurity requirements and Georgia Tech’s own policies’ at the request of its director.”
Georgia Tech disputes the charges. It claims there was no data breach or data leak, the information involved was not confidential anyway, and the government had stated this research did not require cybersecurity restrictions. Really? Then why the (allegedly) falsified cybersecurity score? The suit claims the glowing self-reported score for the Georgia Tech campus:
“… was for a ‘fictitious’ or ‘virtual’ environment and did not apply to any covered contracting system at Georgia Tech that could or would ever process, store or transmit covered defense information.”
That one will be hard to explain away. Other entities with DOD contractor will want to pay attention—Graham states the DOJ is cracking down on contractors that lie about their cyber protections.
Cynthia Murrell, September 4, 2024
Google Synthetic Content Scaffolding
September 3, 2024
This essay is the work of a dumb dinobaby. No smart software required.
Google posted what I think is an important technical paper on the arXiv service. The write up is “Towards Realistic Synthetic User-Generated Content: A Scaffolding Approach to Generating Online Discussions.” The paper has six authors and presumably has the grade of “A”, a mark not award to the stochastic parrot write up about Google-type smart software.
For several years, Google has been exploring ways to make software that would produce content suitable for different use cases. One of these has been an effort to use transformer and other technology to produce synthetic data. The idea is that a set of real data is mimicked by AI so that “real” data does not have to be acquired, intercepted, captured, or scraped from systems in the real-time, highly litigious real world. I am not going to slog through the history of smart software and the research and application of synthetic data. If you are curious, check out Snorkel and the work of the Stanford Artificial Intelligence Lab or SAIL.
The paper I referenced above illustrates that Google is “close” to having a system which can generate allegedly realistic and good enough outputs to simulate the interaction of actual human beings in an online discussion group. I urge you to read the paper, not just the abstract.
Consider this diagram (which I know is impossible to read in this blog format so you will need the PDF of the cited write up):
The important point is that the process for creating synthetic “human” online discussions requires a series of steps. Notice that the final step is “fine tuned.” Why is this important? Most smart software is “tuned” or “calibrated” so that the signals generated by a non-synthetic content set are made to be “close enough” to the synthetic content set. In simpler terms, smart software is steered or shaped to match signals. When the match is “good enough,” the smart software is good enough to be deployed either for a test, a research project, or some use case.
Most of the AI write ups employ steering, directing, massaging, or weaponizing (yes, weaponizing) outputs to achieve an objective. Many jobs will be replaced or supplemented with AI. But the jobs for specialists who can curve fit smart software components to produce “good enough” content to achieve a goal or objective will remain in demand for the foreseeable future.
The paper states in its conclusion:
While these results are promising, this work represents an initial attempt at synthetic discussion thread generation, and there remain numerous avenues for future research. This includes potentially identifying other ways to explicitly encode thread structure, which proved particularly valuable in our results, on top of determining optimal approaches for designing prompts and both the number and type of examples used.
The write up is a preliminary report. It takes months to get data and approvals for this type of public document. How far has Google come between the idea to write up results and this document becoming available on August 15, 2024? My hunch is that Google has come a long way.
What’s the use case for this project? I will let younger, more optimistic minds answer this question. I am a dinobaby, and I have been around long enough to know a potent tool when I encounter one.
Stephen E Arnold, September 3, 2024
Another Big Consulting Firms Does Smart Software… Sort Of
September 3, 2024
This essay is the work of a dumb dinobaby. No smart software required.
Will programmers and developers become targets for prosecution when flaws cripple vital computer systems? That may be a good idea because pointing to the “algorithm” as the cause of a problem does not seem to reduce the number of bugs, glitches, and unintended consequences of software. A write up which itself may be a blend of human and smart software suggests change is afoot.
Thanks, MSFT Copilot. Good enough.
“Judge Rules $400 Million Algorithmic System Illegally Denied Thousands of People’s Medicaid Benefits” reports that software crafted by the services firm Deloitte did not work as the State of Tennessee assumed. Yep, assume. A very interesting word.
The article explains:
The TennCare Connect system—built by Deloitte and other contractors for more than $400 million—is supposed to analyze income and health information to automatically determine eligibility for benefits program applicants. But in practice, the system often doesn’t load the appropriate data, assigns beneficiaries to the wrong households, and makes incorrect eligibility determinations, according to the decision from Middle District of Tennessee Judge Waverly Crenshaw Jr.
At one time, Deloitte was an accounting firm. Then it became a consulting outfit a bit like McKinsey. Well, a lot like that firm and other blue-chip consulting outfits. In its current manifestation, Deloitte is into technology, programming, and smart software. Well, maybe the software is smart but the programmers and the quality control seem to be riding in a different school bus from some other firms’ technical professionals.
The write up points out:
Deloitte was a major beneficiary of the nationwide modernization effort, winning contracts to build automated eligibility systems in more than 20 states, including Tennessee and Texas. Advocacy groups have asked the Federal Trade Commission to investigate Deloitte’s practices in Texas, where they say thousands of residents are similarly being inappropriately denied life-saving benefits by the company’s faulty systems.
In 2016, Cathy O’Neil published Weapons of Math Destruction. Her book had a number of interesting examples of what goes wrong when careless people make assumptions about numerical recipes. If she does another book, she may include this Deloitte case.
Several observations:
- The management methods used to create these smart systems require scrutiny. The downstream consequences are harmful.
- The developers and programmers can be fired, but the failure to have remediating processes in place when something unexpected surfaces must be part of the work process.
- Less informed users and more smart software strikes me as a combustible mixture. When a system ignites, the impacts may reverberate in other smart systems. What entity is going to fix the problem and accept responsibility? The answer is, “No one” unless there are significant consequences.
The State of Tennessee’s experience makes clear that a “brand name”, slick talk, an air of confidence, and possibly ill-informed managers can do harm. The opioid misstep was bad. Now imagine that type of thinking in the form of a fast, indifferent, and flawed “system.” Firing a 25 year old is not the solution.
Stephen E Arnold, September 3, 2024
Consensus: A Gen AI Search Fed on Research, not the Wild Wild Web
September 3, 2024
How does one make an AI search tool that is actually reliable? Maybe start by supplying it with only peer-reviewed papers instead of the whole Internet. Fast Company sings the praises of Consensus in, “Google Who? This New Service Actually Gets AI Search Right.” Writer JR Raphael begins by describing why most AI-powered search engines, including Google, are terrible:
“The problem with most generative AI search services, at the simplest possible level, is that they have no idea what they’re even telling you. By their very nature, the systems that power services like ChatGPT and Gemini simply look at patterns in language without understanding the actual context. And since they include all sorts of random internet rubbish within their source materials, you never know if or how much you can actually trust the info they give you.”
Yep, that pretty much sums it up. So, like us, Raphael was skeptical when he learned of yet another attempt to bring generative AI to search. Once he tried the easy-to-use Consensus, however, he was convinced. He writes:
“In the blink of an eye, Consensus will consult over 200 million scientific research papers and then serve up an ocean of answers for you—with clear context, citations, and even a simple ‘consensus meter’ to show you how much the results vary (because here in the real world, not everything has a simple black-and-white answer!). You can dig deeper into any individual result, too, with helpful features like summarized overviews as well as on-the-fly analyses of each cited study’s quality. Some questions will inevitably result in answers that are more complex than others, but the service does a decent job of trying to simplify as much as possible and put its info into plain English. Consensus provides helpful context on the reliability of every report it mentions.”
See the post for more on using the web-based app, including a few screenshots. Raphael notes that, if one does not have a specific question in mind, the site has long lists of its top answers for curious users to explore. The basic service is free to search with no query cap, but creators hope to entice us with an $8.99/ month premium plan. Of course, this service is not going to help with every type of search. But if the subject is worthy of academic research, Consensus should have the (correct) answers.
Cynthia Murrell, September 3, 2024
-
- Subscribe to Beyond Search
Feature archive
News archive
-
