Google has at least 10x as many employees as it needs. That's based on decades of revenue being decoupled from costs, and from monopoly profits. It's the double-whammy of being in tech (where costs and revenues don't couple since incremental costs are close to zero), and having a super-profitable monopoly.
You can't fire 90% of the workforce; even morale and culture problems aside, at this point, the organization is structured to need too many people. On the other hand, organizing over 100k employees turns into this superorganism problem.
At >100k employees, the influence of anyone -- up to and including the CEO -- is very, very limited. Dynamics take over.
I've been bearish on Google for a long time, but I haven't shorted the stock. Market irrationality can last a lot longer than my wallet. Fundamentally, though, one could muscle through building a new, better Google for much less than the valuation of Google.
Google could also be worth a lot more with a breakup. There is negative synergy between, for example, advertising, Google Workspace, Android, hosting, etc. If those were independent businesses, with partnerships around things like data sharing, the total value could be much higher. Alphabet was a move in the right direction, but the move should have been much deeper. Right now, Alphabet is Google + a bunch of tiny startups.
No one is incentivised to grow head count. They are incentivised to do more, produce more revenue, and expand the reach of the company; most of these goals require the application of more smart people to a problem. They are not getting a promotion or salary bump because their group increased from 200 to 300 people over the previous quarter, they are receiving rewards for being able to effectively direct those people in order to achieve some specific goal. Increasing head count by 50% while delivering 250% growth in revenue/engagement/view/etc is what is being rewarded.
- If I have a job at Google as an IC, I can look for IC jobs.
- If I have a job managing 10 people, I can get a manager-level job.
- If I have a job managing 100 people, I can get a director-level job.
- If I have a job managing 1000 people, I can look for a VP-level job.
- If I have a job managing 10,000 people, I can look for a C-suite job.
Average tenure in a position in the tech industry is around 3 years. No one will know or care what objectives a Google manager achieved, or whether those were hard or easy to achieve. If you want to grow rapidly in your career, the things which matter are:
- Connections, network, and references
- What shows up on an interview (e.g. self-improvement)
- What shows up on your CV (e.g. how many people you managed, or the brands you worked on)
The place techies get stuck in career growth is by trying to do the "right thing." By the time you're trying to rise to the top of a corporate ladder, the competition is extreme, and the people who succeed play the game as optimally as they can.
I might be able to look one level up. An IC might be able to manage a small team. However, your odds of having an IC move directly into the C-suite require nothing short of a miracle (a Nobel prize, being the author of Linux, or something similar).
Actually delivering value is a huge advantage too, but as far as I've seen is not prerequisite for getting ahead.
Appearing to deliver value is strongly advantageous. Delivering value helps you do that. Other methods:
- Good salesmanship (e.g. making easy things look difficult)
- Stealing credit (can be personal, or e.g. for market conditions)
- Lying (e.g. cooking books)
Delivering value is probably the easiest way to appear to deliver value at the IC level. The higher one moves in the corporate ladder, the more important the others become.
In any public company the CEO is just another employee. Maybe he even has less freedom than others.
Yes, responsibility and compensation are vastly greater, but in the end you don't really realize personal goals.
At some point I really hope that we can all recognize that “seemingly harmless”, “engineering-oriented”, and “data-driven” are not virtues.
Perhaps this is the point you are making: if one is data-driven and engineering-oriented without regard to the consequences, that could be seen as a vice.
Despite the amazing engineering/science, it is hard to come to peace with von Braun or Teller (just to pick two examples).
Hopefully we can get to the point of putting purpose (I think “mission” is a bit too hackneyed and neat to work here) first in discussions of culture, whether corporate or engineering culture.
But in hindsight, this was a ploy from them to try and upsell AMP, which was in a lot of cases heavier than non-AMP websites, and it was a means to keep people inside of their ecosystem - and inside of their advertising network, because not all websites Google linked to had Google adverts on it.
They've given up on AMP. They've given up on improving search results quality. Dare I say they've given up on improving the web? I mean they're still putting some money into Chrome and HTTP/3 and the like.
Looking at the misnamed "Privacy Sandbox", Chrome is not just there to improve the Web (and therefore google usage), but also to control it (and steer advertisers to Google even if it undermines the Web as a whole).
They’re killing Firefox with it (RIP) and successfully killed IE with it (rest in piss).
We’re getting to a desktop browser hegemony again but I doubt a scrappy upstart will be able to save us.
I'm hopeful there are a lot of Mac users like me out there.
A listing of top global educational institutions would only have a few thousand entries.
It's not without some irony that I note that the early, and much belated, Internet was largely edu domains, along with a handful of tech firms, government agencies, and military entities.
https://www.goodyear.com/en-US/learn/tire-care-maintenance/h...
On Kagi you have to go until the 10th result to get something from Bridgestone, which explains the same thing while being less aggressive about selling tires:
https://www.bridgestonetire.com/learn/maintenance/how-to-cha...
So the trade-off seems to happen between seeing fewer ads and having better tolerance for imprecise queries.
The Bloomberg Terminal is extremely expensive and yet companies pay for it because it makes their traders' jobs so much more efficient. I don't see why companies wouldn't do the same for a search engine that filters out all the bullshit so their employees don't have to waste their time wading through it.
(Kagi founder here)
I'm a Kagi private beta tester and love it. Will become a paying customer.
One question: I do remember a page that told me how much I searched and what that would mean once Kagi becomes a paid service. I can't find that info anymore. In light of the mail us Kagi users got today, does that mean that it's going to be a flat rate for starters?
We do not yet know if the flat fee will be viable for the company and we hope to learn that very soon.
If you want to save your search history perhaps you can open a suggestion on https://kagifeedback.org
Google has 4.3B users, so 1% of that is 43M. With a $5 subscription, Kagi would have $215M in monthly revenue, enough to pay for 14,000 developer salaries.
A more reasonable target might be 500k users at, let's say, $10/mo.
Maybe not dominate the market, take over the world, but not every thing has to be that.
Imagine a search engine that truly is private, containerized search and does not need to sell you anything. It only wants to serve you true relevance and accuracy. That engine also skips over all monetized sites that serve more than 5-10% ads. Prefers cookieless sites. That search engine would be bliss.
> How we make money
> We plan to fund Andi through a freemium business model, with free anonymous search for everyone for ever, and paid plans for professionals and businesses, including APIs and using Andi for corporate information, with supplemental revenue from anonymous referral link attribution. We're focused on building the product right now, and will figure out the details once we're further along.
> We are committed to staying 100% advertising free. Commercial considerations or partnerships will have zero impact on our search results or recommendations. Our recommendations are always made based on the best results we can find for our customers, and are never subject to commercial influence.
> By sharing search revenue with content producers and media companies that join our network, our mission is to provide funding to great content and reduce media's reliance on invasive ad-tech. Contact us at [email protected] to get involved!
There are only 2 employees[1] of the company so there is a low amount of revenue they need to be self-sufficient
What I really want is a decentralized, non-commercial, open source web. As part of this I would like to see search and communications moved to the client, rather than the cloud. This seems like it would take a huge effort to build, however, and even more monumental effort to bring people into the network. So it feels like a pipe dream at this point. I do think there are a lot of other people out there craving for something similar, based on nostalgia for the old days (90's and earlier).
There's a lot more to search than just search in the traditional sense, and I think multiple cooperating services offering small complimentary functionality slices could probably offer reasonably similar capabilities (save for like image search, maps and translation) with a significantly smaller hardware footprint.
My own search engine is fairly janky, but a lot of its problems are understood and could probably if not be fixed, at least mitigated. That's one functionality slice, searching documents. You could have a slice for forums, one for reviews, one for high value sites like stackoverflow and wikipedia, one for facts, one for popular links in social media and news, one for source code, one for mailing lists, and so on.
As a whole, I think you might actually be able to put together an actually useful information gateway this way. Would be a lot of work, but I think the amount of work probably exceeds its difficulty.
Indeed. The internet in general and the beginnings of the web were a very different place back then. I remember about 1996 or so seeing a URL on a Pepsi can and thinking, "oh crap, it's over."
The only way you have not listed is gov't funding. And i dont think i would trust a gov't run search engine either. May be a well meaning librarian institution could make it work.
> I'm quite sure a subscription model would fail.
this hasn't really been tried yet - it could work, who knows?
"...offers a freemium model (free basic and premium for $5/Month which includes paid versions of VPN....."
Or as it reads?
https://neeva.com/blog/introducing-neeva-free-basic-and-neev...
Do I have to build the startup myself?
Or let Musk buy them.
Every site would have to run their own data analysis and sell their own ads. Advertisers would be unlikely to take Google or anyone else's word, so audits would be necessary but difficult to conduct. It wouldn't completely end online advertising but would turn it back into something closer to the old traditional model where those selling ad space had a handful of semi-generic personality models for their readers/users and advertisers would select one that best matches the profile of their target customer. It destroys the economics of centralized data collection.
There are many other laws that could accomplish a similar end but through different means. The biggest barrier would be getting politicians to not care about the demands of huge tech companies that dump endless amounts of money into lobbying and campaign funding.
I cannot imagine it being easy, but I find it hard to believe that Google aren't able to weed out the worst offenders. For a large number of queries the results are all SEO spam. Try searching for anything related to health and get a useful result. It's no longer possible.
Medicine researching on the internets always was among the hardest topics. And anyone who really understands Medicine has to write on any advice not to consider it as a real medical advice... I agree that google has degraded down to unusable in comparison to non-global searchable websites (like SO) but Medicine is very bad example to your point.
In fact, I wonder how far a search engine could go with a curated list of "good actors" you exclusively serve up results from. Add a "search everywhere" link for people that prefer SEO hell.
When Google started, the web was honest and innocent. People linked to the pages they liked, and Google could crowd source those links and raise up the most popular pages.
Now most of the web is lies meant to confuse search engines. Figuring out what pages are good is no longer a feasible task for software.
Of course it is. It shouldn't be at all difficult to train a large natural language AI model on SEO spam pages. But nobody will do it unless they can figure out a business model that makes it pay.
Or a non-profit organization is formed to do search spidering for the benefit of the whole world, much like Wikipedia or PBS.
At this point, it's no different than an appstore in terms of volume. Ironically that's where we're heading with most websites being "3.0" and just SPAs and now PWAs even.
And who can blame us? Other sites still suck worse. I can't believe how bad walmart.com still is - the user experience is night and day. Walmart - the retailer embracing computerized tracking of spending by credit card number and other advanced data analytics a full decade before Amazon was even a glimmer in Bezo's eye took forever to realize the power of using their stores as local fulfillment centers? Provides a 3rd party marketplace that's less reliable than craigslist? Who the hell is in charge of online operations at these big companies and how do they still have jobs?
Then again just look at Sears - the company that literally invented mail order *in the 1800's* and how they completely missed the Internet - the ultimate mail order environment.
Ugh.
For example, I wanted to watch a bunch of videos of Mark Zuckerberg speaking from before 2010 and the YouTube search engine only showed me a limited amount of things I had already watched— it even had unrelated suggestions below those results based on my viewing habits.
So I went to Google and searched for videos and interviews of Zuck and found a bunch which were not easily surfaced on YouTube itself.
Same thing with Ina Garten/Barefoot Contessa cooking videos.
Now that I've found some fresh videos, the homepage recommendation algorithm has gotten better at surfacing new content related to these topics.
The problem is that Google is engaged in a war with SEO optimizers, and losing (or at least, barely breaking even). No business model will fix that. It's a hard problem that any successful search engine will struggle with.
For many search terms, the entire above-the-fold and sometimes multiple scroll heights are filled with:
- Ads
- Info cards
- People Also Ask
- Map/Business info
- Videos
- Related Products
- More Ads
The organic search results start after that and on mobile they may as well not exist.
If everyone is clicking the directly or indirectly sponsored/algorithmic content then what incentive does Google have to give you good, straightforward links to relevant webpages?
So why would Google serve the more direct and helpful result from the edu domain!
I think they are genuinely struggling in an adversarial environment. Imagine every hacker in the world specifically focused on your system and the payoff is millions or even billions of dollars.
But clearly there are also a lot of motivations to de-emphasize the organic results on the search page in favor of ads, “algorithmic answers” that give you an answer without clicking away from Google, and redirects to Maps, Youtube, and Shopping which are revenue-generators for Google.
Maybe that also comes with a budget de-emphasis for fighting the war in the organic results vs surfacing relevant content from these other sources.
A typical employee at any large company is incentivized to:
- Maximize their salary
- Maximize their future earnings potential (e.g. CV)
- Have fun
- Have work-life balance
... and so on. That goes all the way up to the CEO. Pretending a corporation thinks and responds to incentives is folly. Running a corporation like Google is like herding cats. If everyone marched in the same direction, Google would cut through SEO optimizers like a chainsaw through butter. There is no wayh to do that with 100+k people.
Google fundamentally has no financial incentive to fix any problem, since organizations aren't intelligent, sentient beings which act on incentives. They have dynamics from the interactions of 100k+ individual agents.
There is no magical fairy which makes large organizations act in their own best interests or even try to survive. The only reason they do survive is that's as true of their competitors as themselves.
Corporations are intelligent, sentient entities which act on incentives. They don't have a conscious sense of self, but then neither do many natural organisms. That doesn't keep them from responding to their environment as if they're aware of it, and acting in incentive-driven ways.
Corporations seeking to maximise profits are exactly like humans attempting to maximise income. The problem is that corporations - like humans - have the wrong incentives, further diminished by limited predictive ability and poor heuristics.
So maladaptive behaviours emerge and stick. Both humans and corporations find a nice local minimum and don't move from it.
More intelligent behaviour would be able to predict and avoid existential threats without having to experience them first. But it's a question of predictive ability, not a binary absence of all intelligence.
So it's not that Google can't do better because it has too many employees. It can't do better because its cultural heuristics - which include it internal culture and the external culture it operates in - don't allow it to.
I can set up "incentives" for an anthill, in the form of a mixture of sugar and borax. Emergent behavior takes over.
The behavior of corporations, at the scale of Google, Facebook, or similar, is almost entirely emergent dynamics. Same thing for countries, for that matter.
Neeva also lets me index my personal documents and includes those in search results as well. They are an early startup but I am excited about what they can build when they are not beholden to advertisers.
Google has an impossible task in trying to answer a query like "Portable air mattress reviews", because 99.99999% of the results are SEO-optimized crap
On the other hand, they have an incentive - increased competition, they can still monetize better pages, search is what makes them money.
1 and 2 are trivial to detect. 3 is harder but still not impossible given Google's machine learning capabilities and existing datasets.
Downrank or ban ads & affiliate links and you remove any incentive for creating spam sites in the first place.
This would also put pressure on content creators to tone down the ads or explore alternative monetization models.
I also spend 50£/month on Patreon to support the creators I watch regularly.
For example, all the stupid aliasing of search terms, and dropping of search terms.
This is helpful if you want to optimize for "sorta close, sell things!" such as "maytag A4453 fix timer", but you want advertisers to latch on to maytag=washing machine, and ignore the rest.
And it also helps when websites with google ads on them them get higher SEO, because suggesting that site might get a click, and profit!
EG, suggest a blog/spam site, because google ads are on it.
And it also helps with google voice search! Did the speaker mean here/hear? Did they mean Danny/Daniel/Danson when they said Dan?
Google has too much genericization in its search, and it is for profit naturally, but I think it is lazy profit, and it is becoming worse yearly.
And it is all their fault, no one else's.
Many product folks just don’t seem honest, even with themselves, about what benefits users. Do you really think your users benefit from ”more relevant” advertisements? Because your users don’t. Are new services and content so beneficial to users that they’d forfeit their privacy to fund it? If so, why obfuscate your having made that choice for them? Why not make it opt-in? Does that dialog box popping up just at the right time really give that overstimulated and frustrated user a choice? Either answer those kinds of questions honestly or admit that you’re just finding the smoothest path to maximize revenue. The hypocrisy is infuriating.
Give me a break.
My loosely held theory on the wide variance in people's judgement of DDG is that some people really like/have come to rely on how much Google personalizes results based on all the other data they have on you, even if they are in the "Google has gotten bad" camp. DDG doesn't do any of that: it's much more like going back 20 years where you have to put all the context you care about into a query, you're not searching "in a bubble". For me, that's a very positive feature, but others may find it frustrating and not even be able to articulate why they're not happy with the results. DDG does do a lot of the nice "enhancement" stuff Google pioneered long ago, like embedding the intro of the Wikipedia page in the sidebar for a query with a strong match, e.g. a person's name. So for me DDG is just the right level of "smart": it understands common queries well enough to immediately highlight useful details, but it doesn't try to be so smart that you as a user stop being able to understand what it's doing under the hood.
Also, I think Google could make a ton of money if they sold one ad per language at a time. As in, for the next 5 minutes the ad for all English speaking users will be for coca-cola, and then it will be for boscovs. Just like TV ads.
That is to say, a search engine could sell advertising without betraying it's users or compromising its own quality. It is possible, but would need constant vigilance against the temptations of greed.